Meta Muse vs OpenAI dots

Meta Muse and OpenAI dots are both AI agents with their own computer in the cloud, and as of October 2026 the main differences between them are who can get them and the price to start, where you talk to them, how they handle business use, and how they handle money, approvals and data.^[5][2]

Meta’s Muse is a personal AI agent that Meta launched on September 8, 2026^[2]. OpenAI’s dots are agents in ChatGPT^[7] that OpenAI added on September 29, 2026^[20]. Here is the difference between Meta Muse and OpenAI dots on each of the four points:

• Who can get it and the price to start. Muse is free to start, with a usage limit^[1], and Meta’s September 29 post says Muse is available in the US and Canada^[4]. Your first dot comes with ChatGPT Pro, from $100 a month, or Business Premium^[5][12], and dots are rolling out to Pro users outside the European Economic Area, Switzerland and the UK^[7].
• Where you talk to it. You message Muse in the Muse app or directly in WhatsApp^[2], and the app runs on mobile and Mac^[3]. You message your dot in ChatGPT on desktop and web, and in Slack and Teams^[5], and in the ChatGPT mobile app when mobile access is available^[7].
• Business use. Muse for Small Business adds business skills and connectors inside Muse^[4], and Meta lists no separate price for it^[4]. The ChatGPT Business page says “Dots are available on ChatGPT Business Premium”^[8], and Enterprise workspaces can try a beta once an admin turns it on^[7].
• Money, approvals and data. Muse checks with you before sensitive actions like sending an email or making a purchase, and Meta says a separate Sentinel agent must approve anything that reaches the internet^[2]; you can edit Muse’s Memory.md file directly^[18]. Dots start with built-in rules for when to ask for approval^[5], you can set Custom Rules for some actions^[7], and OpenAI says Auto-review checks certain planned actions before they run^[10]; you currently can’t view or edit individual dot memories^[10].

As of October 2026, here is who should pick Meta Muse and who should pick OpenAI dots: pick Meta Muse if you want a free option and chat in the Muse app or WhatsApp, and pick OpenAI dots if you want a dot included with ChatGPT Pro or Business Premium that you can message in ChatGPT, Slack and Teams.^[5][2][1]

Who should pick Meta Muse, from Meta’s pages:

• Pick Meta Muse if you want a free option to start: Muse is free, with a usage limit.^[1]
• Pick Meta Muse if you live in the US or Canada, where Meta’s September 29 post says Muse is available.^[4]
• Pick Meta Muse if you want to message your agent in the Muse app or directly in WhatsApp.^[2]
• Pick Meta Muse if you run a small business on Meta’s apps: Muse for Small Business can connect Facebook Pages, Instagram professional account analytics and Meta ad accounts.^[4]
• Pick Meta Muse if you want a separate Sentinel agent, which Meta says must approve anything Muse does that reaches the internet^[2], and a memory file you can edit.^[18]

Who should pick OpenAI dots, from OpenAI’s pages:

• Pick OpenAI dots if you have or plan to get ChatGPT Pro, from $100 a month^[12], where your first dot is included at no extra cost.^[5]
• Pick OpenAI dots if you live outside the European Economic Area, Switzerland and the UK, where dots are rolling out to Pro users, or use Business Premium, which has dots across all supported ChatGPT regions.^[7]
• Pick OpenAI dots if you want to message your agent in ChatGPT, Slack and Teams.^[5]
• Pick OpenAI dots if you run a team on ChatGPT Business Premium or Enterprise: dots are available on Business Premium^[8], and an Enterprise admin can turn on a beta.^[7]
• Pick OpenAI dots if you want Auto-review, which OpenAI says checks certain planned actions before they run^[10], and Custom Rules for some actions.^[7]

Who should pick Meta Muse and who should pick OpenAI dots comes down to the point that matters most to you: price and region, where you chat, your business setup, or how each one handles approvals and data.

As of October 2026, Meta Muse costs nothing to start, with a usage limit, and its paid plans cost $20 or $100 a month, while OpenAI dots have no price of their own: your first dot is included at no extra cost in ChatGPT Pro, from $100 a month, or in ChatGPT Business Premium.

How much Meta's Muse costs, per Meta's subscriptions help page^[1]:

• Free: no charge, with a usage limit. When you reach the limit, you can upgrade to a paid plan or wait until your free usage limit refreshes.
• Power plan: $20 a month, with 500M Muse tokens per week.
• Max plan: $100 a month, with 3B Muse tokens per week.

Meta says Muse benefits and availability may vary by region and account. Meta lists no separate price for Muse for Small Business; its September 29 post^[4] links to the standard Muse plans.

Your first OpenAI dot costs what the ChatGPT plan that includes it costs. OpenAI's launch post^[5] says your first dot is included in your Pro or Business Premium plan at no extra cost. The prices of those plans are:

• ChatGPT Pro: Pro 100 at $100, Pro 200 at $200 or Pro 500 at $500 a month, in US dollars, per OpenAI's Pro help page^[6].
• ChatGPT Business Premium: the ChatGPT Business page^[8] says "Dots are available on ChatGPT Business Premium". Separately, OpenAI's Business help page^[14] prices a Premium seat at $125 per user per month when billed monthly, or $100 per user per month when billed annually, and says a Business workspace needs at least 2 Standard or Premium seats in total.

The ChatGPT pricing page^[12] marks dots as not included on ChatGPT Free, Go or Plus.

On both products, a higher price buys more usage. Meta's help page^[1] says the weekly usage allowance for Muse increases as you upgrade plans. OpenAI's Pro help page^[6] says Pro 200 includes more usage than Pro 100, and Pro 500 offers the highest included usage of the three plans. OpenAI's launch post^[5] says conversations with your dot don't count toward your ChatGPT usage limits.

Yes: as of October 2026, Meta Muse is free to use with a usage limit, and when you reach that limit you can either upgrade to a paid plan or wait until the free limit refreshes.

Meta's subscriptions help page^[1] says Muse is available for free with a usage limit. Meta's launch post^[2] puts it in Meta's own words: Muse is "free for most of what people need, with subscription plans for people who want to do more." Meta's business help page^[21] says Muse has a free plan for what most people need, and that paid subscriptions give higher weekly usage limits.

When you hit the free limit, Meta's Muse gives you two choices: upgrade to a paid subscription, or wait until your free usage limit refreshes. Meta's subscriptions help page^[1] says the weekly usage allowance for Muse increases as you upgrade. The two paid plans are:

• Power plan: $20 a month, with 500M Muse tokens per week.
• Max plan: $100 a month, with 3B Muse tokens per week.

Meta's subscriptions help page^[1] gives token amounts for the two paid plans but no amount for the free usage limit. The page also says the benefits you can get are shown during onboarding before you subscribe, and that a paid plan renews automatically each month unless you cancel. To see how much you have used, Meta's business help page^[21] says to open Settings.

Free use of Meta Muse also depends on where you are. Meta's Muse Help Center^[22] says the Muse app and website are not yet available everywhere, and some features may not yet be available in your area. Meta's subscriptions help page^[1] adds that Muse subscription benefits and availability may vary by region and account.

OpenAI dots are not free: as of October 2026, the personal ChatGPT plan that includes them is ChatGPT Pro, from $100 a month, and OpenAI's dots are also included with ChatGPT Business Premium.

The ChatGPT pricing page^[12] lists four personal plans and marks whether each one includes dots:

• Free, $0 a month: dots not included.
• Go, $8 a month: dots not included.
• Plus, $20 a month: dots not included.
• Pro, from $100 a month: the plan card lists "Dot, your always-on agent" and "Your choice of 3 usage tiers".

A dot does not cost extra on top of the plan. OpenAI's launch post^[5] says your first dot is included in your Pro or Business Premium plan at no extra cost, and your plan also includes an allowance for deeper work. The same post says conversations with your dot don't count toward your ChatGPT usage limits, while tasks your dot starts or manages in Codex or ChatGPT Work count toward your usage limits as usual. The post also says that in the future you'll be able to add more dots; as of October 2026 that is announced, not live.

Outside personal plans, the ChatGPT Business page^[8] says "Dots are available on ChatGPT Business Premium". OpenAI's release notes^[20] say dots are rolling out gradually to eligible Pro and Business Premium users aged 18 and older in supported markets, and that Enterprise access is a beta and is off by default.

The lowest ChatGPT Pro tier you need for a dot is Pro 100, and Pro 200 and Pro 500 also work: as of October 2026, OpenAI's dots documentation lists OpenAI dots for Pro 100, Pro 200 and Pro 500 alike.

Choosing between Pro 100, Pro 200 and Pro 500 changes the price, the included usage and Ultrafast. OpenAI's Pro help page^[6] compares the three Pro tiers:

• Pro 100: $100 a month. No Ultrafast. Less included usage than Pro 200.
• Pro 200: $200 a month. No Ultrafast. More included usage than Pro 100.
• Pro 500: $500 a month. Ultrafast included. The highest included usage of the three tiers.

Among Pro plans, Ultrafast is available only on Pro 500, which includes Astra Ultrafast. Pro 200 is available for new subscriptions again. A new Pro 200 subscription that isn't eligible for grandfathering includes a lower usage allowance than previously offered, and the monthly price of Pro 200 stays $200, per the same help page^[6].

All three Pro tiers are billed monthly; annual billing and payment for several months in advance are not available. The ChatGPT pricing page^[12] shows "Your choice of 3 usage tiers" on the Pro plan card. OpenAI's dots docs^[23] put Pro 100, Pro 200 and Pro 500 on one access line for users over 18 outside the European Economic Area, the United Kingdom and Switzerland.

As of October 2026, Meta Muse bills its paid plans monthly, through the mobile app or at muse.ai, while the ChatGPT plans that include OpenAI dots bill Pro monthly and let ChatGPT Business choose monthly or annual billing.

Meta's Muse subscription renews automatically each month unless you cancel, and Meta's subscriptions help page^[1] says you're charged on the same date each month as your original purchase date. The same page says you can subscribe with your Meta account through the mobile app or at muse.ai. Meta's general billing page^[24] for subscriptions from Meta lists payment methods by store: the Apple App Store and the Google Play Store. The same page says that if your billing date doesn't exist in a month, such as the 31st in a 30-day month, you're charged on that month's last day. If your payment fails and you have no backup payment method, you may get a short grace period to fix it before the subscription is canceled. Meta's manage page^[25] says you manage the subscription from the same platform, device type and account you first used, and that a payment method that expires or fails may stop the subscription from renewing.

For ChatGPT Pro, the personal plan that includes dots, OpenAI's Pro help page^[6] says Pro is billed monthly, and annual billing is not available. The ChatGPT pricing page^[12] says you can buy Go, Plus, Pro or Business with any major credit card, and that Enterprise customers contact sales for options such as invoicing. For ChatGPT Business, where dots are available on Business Premium, OpenAI's sign-up page^[26] says you choose monthly billing or an annual plan at checkout, and a new account needs at least two paid seats. OpenAI's Business billing page^[27] says Business supports credit and debit card payments. Adding paid seats brings an immediate prorated charge, and seat reductions take effect at the next billing cycle, per OpenAI's Business FAQ^[28].

Two rules from OpenAI's Terms of Use^[29] apply to all of ChatGPT: paid subscriptions renew automatically until you cancel, and OpenAI gives at least 30 days' notice before a subscription price increase, which then takes effect on your next renewal.

As of October 2026, canceling Meta Muse keeps your paid benefits until the billing cycle ends, and canceling ChatGPT Pro, the personal plan that includes OpenAI dots, normally keeps your plan until the billing period ends; neither cancellation brings a refund by itself.

Meta's cancellation help page^[30] says that when you cancel your Muse subscription, you keep access to all subscription benefits until the end of your current billing cycle. You won't receive a credit or partial refund unless required by law. If you bought through the Apple App Store or Google Play Store, you follow that store's cancellation process. Signing out of your account or deleting the Muse app doesn't cancel the subscription. In the Muse app or on muse.ai, you open Settings, then Manage subscription, then select Cancel and Yes, cancel. You can subscribe again at any time. Meta's payment help page^[31] for subscriptions from Meta says to cancel at least 24 hours before your next payment date to avoid a charge for the next billing cycle.

For ChatGPT Pro, the personal plan that includes dots, OpenAI's Pro help page^[6] says to cancel through the platform where you bought the subscription: ChatGPT, Apple or Google Play. Canceling stops future renewals, and you can normally keep using your plan until the end of the current billing period. Uninstalling the ChatGPT app does not cancel your subscription. A scheduled cancellation doesn't end the subscription at once, and you can undo it before the billing period ends. Refund eligibility depends on your purchase and applicable rights, and canceling does not automatically refund charges you have already paid. For Apple App Store purchases, you request the refund from Apple.

OpenAI's Terms of Use^[29], which apply to all of ChatGPT, say payments are non-refundable, except where required by law.

As of October 2026, Meta Muse is available in the US and Canada, and OpenAI dots are rolling out to ChatGPT Pro users in markets outside the European Economic Area, Switzerland and the UK, and to Business Premium users across all supported ChatGPT regions, according to OpenAI's help center.

For Meta Muse, Meta's September 29 post^[4] describes Muse as "a personal AI agent available in the US and Canada." Meta's Muse first rolled out in the US on September 8, 2026, on iOS, Android and muse.ai, per Meta's launch post^[2]. Meta's business help page^[21] says Muse is currently available to people age 18 or older in the United States and Canada. The muse.ai business page^[32] says Meta is working to bring Muse to more markets, with a measured approach in each one, and has no additional timing to share. Meta's subscriptions help page^[1] says Muse and Muse subscriptions are in limited testing and aren't available in all locations yet.

For OpenAI dots, OpenAI's help center^[7] says dots are rolling out to Pro users "in markets excluding the European Economic Area, Switzerland, and the UK," and are available to Business Premium users "across all supported ChatGPT regions." Other OpenAI pages word it differently:

• OpenAI's release notes^[20]: eligible Pro and Business Premium users "in supported markets," with Pro access excluding the European Economic Area, Switzerland and the UK "at launch."
• The ChatGPT Business page^[8]: "Available to Business Premium users in eligible markets."
• OpenAI's dots docs^[23]: Pro 100, Pro 200 and Pro 500 "outside the European Economic Area, United Kingdom, and Switzerland," and Business Premium and Enterprise "Rolling out worldwide."

For people in the UK, Switzerland or the European Economic Area, Meta's pages name the US and Canada for Meta Muse, and OpenAI's pages leave those places out of OpenAI dots on Pro, while the help center says Business Premium users have dots across all supported ChatGPT regions. The help center also says dots are rolling out gradually, so access may take several days to reach your account.

As of October 2026, neither Meta Muse nor OpenAI dots is open to people under 18, and Meta's terms add a higher age where a country or territory requires one.

The Muse Supplemental Terms^[13] say you cannot use Meta's Muse if you are under the age of 18 "(or such greater age required in your country or territory)." Meta's subscriptions help page^[1] says you must be 18 or the age of majority in your country to subscribe. Meta's business help page^[21] says Muse is currently available to people age 18 or older in the United States and Canada, and the muse.ai business page^[32] says it is available to people in the US and Canada "who are 18 and over."

OpenAI's dots safety FAQ^[10] answers the question directly: "No. Dots are not yet available to users under 18." OpenAI's release notes^[20] say dots are rolling out to eligible Pro and Business Premium users "aged 18 and older." OpenAI's dots docs^[23] word the Pro rule as "For users over 18" outside the European Economic Area, the United Kingdom and Switzerland.

A wider rule applies to all of ChatGPT. OpenAI's Terms of Use^[29] say you must be at least 13 years old, or the minimum age required in your country, and that users under 18 need a parent's or legal guardian's permission. For OpenAI dots, the safety FAQ's under-18 answer still applies.

As of October 2026, Meta Muse needs a Meta account, and OpenAI dots need a ChatGPT account on ChatGPT Pro or ChatGPT Business Premium, or in an Enterprise workspace where an admin has turned dots on.

Meta's getting-started page^[33] says that whether you use Meta's Muse from the app or the web, you connect with your Meta account. If you don't have a Meta account, or don't want to use an existing one, you can create one with a Facebook or Instagram account, an email address or a phone number. Meta's subscriptions help page^[1] says you subscribe with your Meta account through the mobile app or at muse.ai. The Muse Supplemental Terms^[13] say you can keep Muse in a separate Accounts Center by signing up with an email address that is not associated with another Meta account. If your Facebook, Instagram and Threads accounts are in the same Accounts Center, Meta's connectors page^[17] says those three connect to Muse automatically. Meta may use your account email or phone number for notifications, including two-factor authentication and alerts about unrecognized logins, per Meta's account help page^[34].

OpenAI's dots live inside ChatGPT. The ChatGPT pricing page^[12] lists dots on the Pro plan and not on Free, Go or Plus, and the ChatGPT Business page^[8] says "Dots are available on ChatGPT Business Premium". A new Business account must add at least two paid seats, and the person who signs up gets the owner role, per OpenAI's sign-up page^[26]. OpenAI's help center^[7] says Enterprise users, including Edu and Healthcare, can try the dots beta when their workspace admin enables it. Once you have access, you create your dot in the ChatGPT desktop app or in ChatGPT on desktop web, and your dot receives memories from ChatGPT.

As of October 2026, you talk to Meta Muse in the Muse app on iOS, Android and Mac, on the web at muse.ai and in WhatsApp, and you talk to OpenAI dots in ChatGPT and in Slack and Microsoft Teams.

Meta's launch post^[2] says talking to Meta's Muse works like messaging another person, in the Muse app or directly in WhatsApp, and that Muse is rolling out in the US on iOS, Android and muse.ai. The muse.ai site^[3] says you chat with your agent using the Muse app on mobile, Mac, or directly in WhatsApp. Meta's business help page^[21] lists muse.ai and the Muse app on macOS, iOS and Android. Meta's security page^[16] names the Muse clients as the iOS and Android apps and the web UI.

OpenAI's launch post^[5] says you can message or call dots in ChatGPT on desktop, web and mobile, and that you can also message your dot in Slack and Teams. OpenAI's help center^[7] gives the setup details. You create a dot in the ChatGPT desktop app or in ChatGPT on desktop web, and the desktop app is also available on Windows. You cannot currently create a dot on mobile, and dots are not supported on mobile web. After you create your dot, you can talk to it in the ChatGPT mobile app when mobile access is available. A dot cannot start a call to you at launch.

OpenAI's pages differ on texting. OpenAI's help center^[7] describes texting as a limited beta for Pro users in the US. OpenAI's launch post^[5] and OpenAI's dots docs^[35] describe texting as coming soon.

As of October 2026, you can sign in to and use Meta Muse in its iOS and Android apps and also message Muse in WhatsApp, while OpenAI dots must be created on a desktop first and then used in the ChatGPT mobile app when mobile access is available.

Meta's Muse rolled out on iOS, Android and muse.ai, per Meta's launch post^[2], and the same post says you can message Muse in the Muse app or directly in WhatsApp. Meta's security page^[16] lists the iOS and Android apps among the Muse clients. Meta's getting-started page^[33] says you connect with your Meta account from the app or the web, and you can create a Meta account with Facebook, Instagram, an email address or a phone number. Meta's subscribe page^[36] gives the steps to subscribe in the mobile app: open Menu, then Settings, tap Upgrade under Usage, select a plan, then continue to payment.

OpenAI's help center^[7] says you cannot currently create a dot on mobile, and dots are not supported on mobile web. You create your dot in the ChatGPT desktop app or in ChatGPT on desktop web, and you use desktop to connect messaging channels, such as Slack or texting. OpenAI's launch post^[5] says that after the initial setup, you'll be able to message your dot in the ChatGPT mobile app. OpenAI's dots docs^[23] add that mobile use needs the supporting app update. ChatGPT has iOS and Android apps, per OpenAI's download page^[37].

Once OpenAI dots work in your mobile app, OpenAI's help center^[7] says you can review or connect apps from your dot's profile under Customize, then Plugins, and that on mobile the dot's computer opens in your control. Texting is a separate phone channel: the help center describes it as a limited beta for Pro users in the US, and if texting is available for your account, you connect your phone from the desktop app.

As of October 2026, you connect OpenAI dots to Slack or Microsoft Teams from your dot's profile, and then message your dot there; in Slack you can send a direct message or mention your dot in a channel thread.

OpenAI's launch post^[5] says you can message your dot in Slack and Teams. OpenAI's dots docs^[35] say to open your dot's profile and select Add to see and connect the contact methods available to you. OpenAI's help center^[7] says to set up messaging channels from desktop.

OpenAI's channel docs^[35] say that in Slack, you can send a direct message or add your dot to a channel and mention it in a thread. By default, your dot responds to you in Slack, and you can instruct your dot to engage with others. The docs also say existing permissions and safeguards still apply to what your dot shares and does, and your dot can message you privately to ask what you're comfortable sharing before it replies in a channel. Adding your dot to a channel doesn't set up a monitoring schedule. For monitoring, you tell your dot which channels to check, what to watch for and when to notify you, then ask it to confirm the recurring task.

You reach the same dot in ChatGPT, Slack and Teams, and changing where you talk to your dot doesn't start a new dot or reset its memory, per OpenAI's dots docs^[23]. OpenAI's channel docs^[35] say Slack and Teams conversations show the messages exchanged there and don't mirror every message from other channels. You can tell your dot where you want different updates, for example routine progress in ChatGPT and decisions in Slack. Connecting a messaging channel doesn't automatically give your dot access to an inbox, other apps or your computer, per OpenAI's getting-started docs^[38].

In Enterprise workspaces, OpenAI's workspace help page^[15] lists an admin setting, Add dots to Slack and Microsoft Teams. The setting allows a dot to join supported Slack or Microsoft Teams workspaces and post with its own identity, where available. Members must complete setup after the permission is enabled.

Yes, in a limited beta: as of October 2026, OpenAI's help center says texting OpenAI dots is a limited beta for Pro users in the US, while other OpenAI pages describe texting your dot as coming soon.

OpenAI's help center^[7] says that where available, texting uses a third-party provider and is offered as a limited beta. The texting beta is limited to Pro users in the US and is not available in Business or Enterprise workspaces. Access is limited and may not be available to every Pro user, so a US Pro plan alone doesn't guarantee you can text your dot.

Other OpenAI pages describe texting differently. OpenAI's launch post^[5] says you can message your dot in Slack and Teams, "with texting coming soon." The ChatGPT dots feature page^[9] says "Texting your dot is coming next." OpenAI's dots docs^[35] say "Texting is coming soon."

If texting is available for your account, OpenAI's help center^[7] says you use the desktop app to connect your phone and follow the setup prompts. After setup, you can message your dot from your phone. The help center also says to exercise caution when sharing sensitive information through text messages.

To stop texts, OpenAI's help center^[7] says to reply STOP to stop outgoing text messages from your dot. On cost, the help center says message and data rates may apply, and asks you to review the notice when you enable texting, along with OpenAI's Terms of Use and Privacy Policy.

Yes: as of October 2026, Meta says Meta Muse stays aware across its main chat and side chats and uses context from the apps you connect, and OpenAI says OpenAI dots can use relevant context across ChatGPT, Slack, Teams and connected apps.

Meta's Muse is built around one long conversation in a main chat. Meta's design post^[19] says Muse's memory persists across your conversations and can manage lots of context, and that Meta built side chats for people who wanted separate context for some topics. The same post says you can interrupt Muse or send multiple tasks at once. Meta's getting-started page^[33] says side chats are separate conversations, and Muse stays aware across the main chat and side chats. Muse is not turn-by-turn, so you can follow up naturally without repeating the context. Meta's personalization page^[39] says Muse's understanding of you comes from what you share in conversations, patterns it observes, context from connectors you've connected, and its file-based memory.

OpenAI's dots docs^[23] say you reach the same dot in ChatGPT, Slack, Teams or a call. Changing where you talk to your dot doesn't start a new dot or reset its memory. Messages stay in their own channels, while your dot can use relevant context across them. OpenAI's channel docs^[35] add that Slack and Teams conversations don't mirror every message from other channels, and that before sharing information from a private conversation with other people, your dot checks that you've allowed it. OpenAI's help center^[7] says your dot receives memories from ChatGPT and can create its own memories, including from connected apps. A messaging channel is a separate connection: connecting one doesn't automatically give your dot access to an inbox, other apps or your computer, per OpenAI's getting-started docs^[38].

As of October 2026, Meta Muse handles business use through Muse for Small Business, a set of business skills and connectors inside Muse, while OpenAI dots are available on ChatGPT Business Premium and, as a beta, in ChatGPT Enterprise workspaces.^[7][8][4]

Meta announced Muse for Small Business on September 29, 2026, as “a collection of new skills and connectors inside Muse to help people run their businesses”. Meta says it built the offering for small business owners “to help get work done with the tools they already use”. Muse for Small Business connects Instagram professional account analytics, Facebook Pages and Meta ad accounts, and, in Meta’s words, “dozens of tools that businesses already run on”. Meta lists no separate price for Muse for Small Business; its post links to the standard Muse plans.^[4] Meta’s business help page says Muse can get things done “at home and now at work too”.^[21] Meta’s pages don’t mention team seats or admin controls for Muse.

OpenAI’s ChatGPT Business page says “Dots are available on ChatGPT Business Premium”.^[8] OpenAI’s Business help page lists the Premium seat at “$125 per user per month when billed monthly” or “$100 per user per month when billed annually”, and says a Business workspace “requires at least 2 Standard or Premium seats in total”.^[14] Enterprise users, including Edu and Healthcare, can try the dots beta when their workspace admin enables it; it is turned off by default.^[7] OpenAI’s help page for Enterprise workspaces lists admin settings for dots, such as whether members can use dots and whether a dot can join Slack or Microsoft Teams.^[15] In OpenAI’s words, dots can “connect to over 4,000 apps” through plugins.^[5]

Pick by how your business runs: Meta business accounts and tools connected inside Muse, or dots in a ChatGPT Business Premium or Enterprise workspace.

Muse for Small Business is a collection of new business skills and connectors that Meta added inside Meta Muse, Meta’s personal AI agent, and announced on September 29, 2026.^[2][4]

Meta says it built Muse for Small Business for small businesses, “to help get work done with the tools they already use”.^[4] Meta’s business help page says Muse can get things done for you “at home and now at work too”, and that Muse can use information from Facebook Pages, Instagram professional accounts and Meta ad accounts alongside the other services you connect.^[21]

Meta describes what Muse does for a business this way: “Suggest ideas based on your business, turn goals into plans and complete multi-step or scheduled work across services you connect.” Meta adds that Muse can keep working after you close it, works in its own isolated space, and checks with you before you publish, send, spend or share.^[21] Meta’s post says Muse will also come to you with ideas, such as flagging emails that need a response and writing your first drafts.^[4]

Meta’s business help page sorts the requests a business can give Muse into five areas: understanding and growing your business, saving time and managing operations, creating content and ads to get new customers, communicating with customers, and analyzing business performance.^[21]

As of October 2026, Meta’s business help page says “Muse is currently available to people age 18 or older in the United States and Canada”.^[21] Meta lists no separate price for Muse for Small Business; its post links to the standard Muse plans.^[4] Meta’s pages don’t mention team seats or admin controls for Muse.

As of October 2026, Muse for Small Business has no price of its own: Meta lists no separate price for it, and its post links to the standard Meta Muse plans.^[4]

Meta’s business help page says “Muse has a free plan for what most people need” and “Power and Maximum subscriptions are available for higher weekly usage limits”.^[21] On its Muse for Small Business page, Meta answers “What does it cost?” this way: Muse is free for most of what people need, with subscription plans for those who want more.^[32]

Meta’s subscription help page lists the two paid Muse plans:

• Power plan: $20 a month, with 500M Muse tokens per week.^[1]
• Max plan: $100 a month, with 3B Muse tokens per week.^[1]

Paid plans renew automatically each month unless you cancel, and Meta says subscription benefits and availability may vary by region and account. You can subscribe with your Meta account through the mobile app or at muse.ai.^[1] Meta’s billing help page says the payment methods you can use depend on where you subscribe, and lists the Apple App Store and Google Play Store.^[24]

Meta’s pages don’t mention team seats or admin controls for Muse.

As of October 2026, Muse for Small Business, Meta’s set of business skills and connectors inside Meta Muse, connects to your Facebook and Instagram business accounts, your Meta ad accounts, and business tools that Meta names: Asana, Box, Canva, Dropbox, Figma, Granola, HighLevel, Intuit QuickBooks, Klaviyo, Lovable, Notion, Shopify, Slack, Stripe and Zoom. Meta says the full list of available connectors is in your Muse app settings, “with more to come”.^[4]

For Meta’s own apps, Meta’s post says Muse can connect your Instagram professional account analytics, Facebook Pages and Meta ad accounts “in a few clicks”.^[4] Meta’s muse.ai/business page says Muse “connects your Facebook, Instagram and Meta ads in one step”.^[32] Meta’s Connectors help page says Facebook, Instagram and Threads are connected automatically if your accounts are in the same Accounts Center, so you don’t need to set those up.^[17]

The named business tools are part of a longer list. In Meta’s words, Muse “can also connect to dozens of tools that businesses already run on”.^[4] The muse.ai/business page ends the same list of names with “and more”.^[32] Meta’s business help page says available connectors may vary by account, and to check Settings for the current list.^[21]

For a service that isn’t listed, Meta’s business help page says you can set up a connection yourself if the service allows it, or let Muse use the website in its own browser.^[21]

As of October 2026, Meta’s business help page describes Meta Muse as Meta’s agent for goals and multi-step work, Meta AI as help with the task you are working on now, and Meta Business Agent as a tool that messages your customers.^[21]

Meta says it “offers different AI tools for different business needs, and you may use more than one”. For Muse, Meta’s description is: “Suggest ideas based on your business, turn goals into plans and complete multi-step or scheduled work across services you connect. Muse can keep working after you close it.” Meta adds that Muse works in its own isolated space and checks with you before you publish, send, spend or share. Muse is at muse.ai and in the Muse app on macOS, iOS and Android.^[21]

Meta AI is for the task in front of you. Meta says Meta AI helps you ask questions, create content and understand business information you share or connect, including for your Instagram, Facebook and Meta ad accounts. Meta AI is in the Meta AI app and at meta.ai.^[21]

Meta Business Agent is for talking with customers. Meta’s description is: “Message customers, capture leads, book appointments and support sales. Available on WhatsApp, Messenger and Instagram Direct (coming soon).”^[21]

Meta lists a fourth tool, Meta AI Business Assistant, for analyzing Meta ad campaigns, fixing common account issues and getting recommendations, in Meta’s advertising tools, such as Ads Manager. Meta notes that features and availability may vary by account, location and device.^[21]

As of October 2026, OpenAI dots are always-on agents in ChatGPT that keep making progress between conversations, while ChatGPT workspace agents and @ChatGPT in Slack are other OpenAI features for Business and Enterprise plans, each described on its own OpenAI page.^[41][40][7]

OpenAI’s help center says your dot “can take on ongoing responsibility and keep making progress between conversations”.^[7] OpenAI’s docs say that in Slack you can send your dot a direct message or add it to a channel and mention it in a thread, and that by default it responds to you.^[35] OpenAI’s safety FAQ says you can set up a separate Slack account for your dot, giving it its own identity.^[10] OpenAI’s dots page says access is available across Pro, Business Premium and Enterprise plans in eligible markets.^[9]

ChatGPT workspace agents let you build agents for repeatable tasks and workflows in ChatGPT. OpenAI’s help page says you can share them with teammates or your workspace, use them in Slack, run them on a schedule, or trigger them through an API.^[40] The ChatGPT Agents app brings ChatGPT workspace agents to Slack, and it needs a Business, Edu or Enterprise ChatGPT account.^[42]

@ChatGPT is for team conversations in Slack and Microsoft Teams. OpenAI’s DevDay recap says you mention @ChatGPT in a channel, thread or DM “to get help with your team’s work”, and that teammates can add context in the same conversation without needing an individual ChatGPT license. @ChatGPT is “Available to all Business and Enterprise plans.”^[41]

OpenAI’s help pages for workspace agents and for the ChatGPT Agents app don’t mention dots.^[42][40]

Put simply: your dot is your own ongoing agent, a workspace agent is built for a repeatable workflow and can be shared with your team, and @ChatGPT helps when someone mentions it in a team conversation.

As of October 2026, Meta says Meta Muse does not publish, send or spend without your approval, and that includes Muse for Small Business, which can connect to your Instagram, Facebook and Meta ad accounts.^[4]

Meta’s Muse for Small Business post says: “You’re in control: nothing publishes, sends, or spends without your approval.”^[4] Meta’s business help page says Muse “checks with you before you publish, send, spend or share”, and that “Muse is designed to check before you spend, send or share. Review the request when Muse asks for approval.” One of Meta’s suggested business requests asks Muse to draft a week of posts and plan an ad campaign “for my review”.^[21]

Meta’s muse.ai/business page asks “Will it post or spend money without asking?” and answers: “No. Nothing publishes, sends or spends without your approval, and you can undo it.”^[32] Meta’s help page on approvals says you can ask Muse to undo or stop some actions in chat, and adds: “Some other actions, like sending an email, cannot be reversed, just like if you were doing them yourself.”^[43]

You can also choose how often Muse asks. In Settings, under Permissions, the connector setting “Ask for some actions” asks before every write action and important read actions, and “Always ask” asks before any action. You can also manage permissions per connector.^[43]

As of October 2026, OpenAI dots are available on ChatGPT Business Premium, according to OpenAI’s ChatGPT Business page,^[8] and OpenAI’s Business help page lists the Premium seat at “$125 per user per month when billed monthly” or “$100 per user per month when billed annually”.^[14]

The ChatGPT Business page’s exact words are “Dots are available on ChatGPT Business Premium”.^[8] OpenAI’s launch post says your first dot is included in your Pro or Business Premium plan at no extra cost.^[5]

A Business workspace “requires at least 2 Standard or Premium seats in total: 2 of either type, or 1 of each”. The other seat type, Standard, is listed at $25 per user per month on the monthly plan, or $20 per user per month billed annually.^[14] When you create a new Business account, you must add at least two paid seats and pick monthly billing or an annual plan at checkout, and the person who signs up gets the owner role.^[26] OpenAI’s Business FAQ says prices are in USD and may vary by country and currency.^[28]

OpenAI says Premium seats include 5x more usage than Standard seats and are not subject to a five-hour usage limit. Buying Premium capacity does not automatically assign it to every member.^[28] An owner can upgrade a member’s Standard seat to Premium by paying the prorated difference for the rest of the billing cycle.^[27]

On where dots work, each OpenAI page uses its own words. The ChatGPT Business page says dots are available to Business Premium users “in eligible markets”.^[8] OpenAI’s help center says dots are available to Business Premium users “across all supported ChatGPT regions”.^[7]

Yes, as a beta: as of October 2026, ChatGPT Enterprise users, including Edu and Healthcare, can try OpenAI dots when their workspace admin enables them, and dots start out turned off by default.^[7]

OpenAI’s DevDay recap says the same: Enterprise, Edu and Healthcare users can try the beta when their workspace admin enables it, and it is off by default.^[41] OpenAI’s docs list Enterprise as “Rolling out worldwide” and say dots “are off by default and must be enabled by a workspace administrator”.^[23] In the workspace settings, the “Use dots (Beta)” permission is “Off by default for Enterprise”.^[15]

OpenAI’s enterprise docs list limits for the beta. During the Enterprise beta, dots do not support data residency or inference residency, and eligible workspaces can opt in after acknowledging those limits. Dots are unavailable for FedRAMP workspaces, workspaces with EKM, and workspaces with inference residency set to the UAE. HIPAA workspaces can take part if they meet the other eligibility requirements. The same docs say neither dots nor Work with local computer access provides strict zero data retention.^[44]

Other dots settings also start off for Enterprise. Custom rules for dots are off by default, and default action rules still apply when custom rules are unavailable.^[15] For records, OpenAI’s docs point Enterprise admins to the Analytics API for dots usage and the Compliance API for supported audit records.^[44]

As of October 2026, Meta’s pages don’t mention team seats or admin controls for Meta Muse, including Muse for Small Business, while OpenAI’s help page for Enterprise workspaces lists admin settings for OpenAI dots, such as who can use dots, Slack and Teams access, local computer access and custom rules.^[15]

The controls Meta’s help pages describe for Meta Muse are settings in your own Muse. In Settings, under Permissions, the connector setting “Ask for some actions” asks before every write action and important read actions, and “Always ask” asks before any action. You can manage permissions per connector, and approvals for each scheduled task and artifact.^[43] Many connectors can be set so Muse can only retrieve data, not take actions.^[17] Meta’s business help page asks you to use Muse only with information and accounts you are allowed to access.^[21]

OpenAI’s help page “Manage dots in ChatGPT workspaces” is written for Enterprise workspaces and lists these settings for dots:^[15]

• Use dots (Beta): lets members use dots; off by default for Enterprise.^[15]
• Add dots to Slack and Microsoft Teams: lets a dot join supported Slack or Teams workspaces and post with its own identity, where available.^[15]
• Allow local computer access: lets a dot use a member’s local files and run commands on their computer; off by default for Enterprise.^[15]
• Use custom rules for dots: lets members add or edit rules that guide their dot’s actions and confirmations; off by default for Enterprise.^[15]
• Cloud computer controls: cloud browser use, cloud network access and cloud computer use, plus a separate password manager setting, all of which apply to dots even when Work is disabled.^[15]

OpenAI’s docs say that if a workspace disables custom rules, saved rules can’t be edited and don’t apply,^[45] and the help page says default action rules still apply.^[15] The help page does not say these settings exist in Business Premium.

Specialist dots are a version of OpenAI dots that OpenAI has announced as a preview for organizations, and OpenAI says it is working with Microsoft to bring them to Agent 365; as of October 2026, both are announced and not live.^[5]

OpenAI’s launch post says your own dot works on your behalf, while specialist dots “take on dedicated responsibilities within your organization”. Your company sets up each specialist dot with its own identity, credentials and access to the systems it needs. The preview covers specialist dots “with their own identity for access management, IT-provisioned hardware, and support for deep integrations with a company’s systems of record”.^[5]

OpenAI says it is building on early testing inside OpenAI across procurement, invoice processing, email marketing, customer support and commercial contracting. OpenAI also says: “We’re starting with focused enterprise pilots.” OpenAI’s engineering teams will work directly with organizations to define each dot’s responsibilities, the tools it can use, and how people review and approve its work. OpenAI asks organizations with work they would like a specialist dot to take on to get in touch.^[5]

On Microsoft, OpenAI says: “We’re also working with Microsoft to integrate specialist dots with their enterprise governance and security controls in Agent 365.” The stated goal is to let businesses manage dots through the Microsoft tools they already use. Both specialist dots and the Agent 365 integration are announced, not live, as of October 2026.^[5]

As of October 2026, Meta Muse can connect to “dozens” of business tools, in Meta’s words, with no exact number published, and OpenAI dots can connect to “over 4,000 apps”, in OpenAI’s words.^[5][4]

For Meta Muse, Meta’s Muse for Small Business post says Muse “can also connect to dozens of tools that businesses already run on”, and that the full list of available connectors is in your Muse app settings, “with more to come”. The post names tools such as Asana, Shopify, Slack and Stripe.^[4] Meta’s Connect 2026 post says Muse “launched with dozens of partners along with access to the entire Shopify catalogue”, and names Notion, Granola, GitHub and Box for getting work done.^[46]

For OpenAI dots, OpenAI’s launch post says dots can “readily connect to over 4,000 apps” through its “ecosystem of plugins”.^[5] OpenAI’s docs give Gmail, Google Drive and GitHub as examples, and say you install and connect each plugin you want your dot to use.^[47] OpenAI’s help center says plugin connections are shared across dots, ChatGPT, ChatGPT Work and Codex.^[10]

Both numbers are each vendor’s own figure, not a count this site checked. Both vendors also say the list you see can vary: Meta’s business help page says available connectors may vary by account,^[21] and OpenAI’s help center says available apps may differ.^[7]

As of October 2026, you connect apps to Meta Muse by asking Muse or from Settings, then Connectors, and you connect apps to OpenAI dots as plugins in ChatGPT’s Plugins tab, where connections are shared across dots, ChatGPT, ChatGPT Work and Codex.^[10][17]

Meta’s help page says you can ask Muse, for example “Connect my Gmail”, or go to Settings, tap Connectors, tap Connect next to the one you want, tap Continue to authorize, and follow any other login steps. Facebook, Instagram and Threads connect automatically if your accounts are in the same Accounts Center. Many connectors can be set so that Muse can only retrieve data, not take actions such as sending a calendar invite. Connectors have their own terms and privacy policies.^[17] Meta says Muse exchanges only the data a connector needs: connecting your email lets Muse search it, but it won’t download your whole inbox unless you tell it to. Meta flags connectors that share information with Muse on their own before you connect them.^[48]

OpenAI’s help center says you connect new plugins and manage existing ones in ChatGPT’s Plugins tab.^[10] On mobile, you open your dot’s profile, go to Customize, then Plugins, and review an app’s permissions before connecting it.^[7] Your existing ChatGPT app permissions apply to your dot, so you can let it read email without letting it send email. If a connection expires, you open the request to reconnect it.^[47] You can also sign into a website in your dot’s browser, using secure sign-in where supported.^[11] A rule that applies to all of ChatGPT: every plugin discloses the data it can access, the actions it can take and who built it, and you review these permissions before anything is shared.^[49]

Both vendors say you choose which apps the agent can use. Meta says people choose which apps Muse connects to and how much access it gets.^[2] OpenAI says you choose which apps dots can access and manage their permissions through your existing ChatGPT app controls.^[5]

Yes, as of October 2026 Meta Muse and OpenAI dots can both work with an app that isn’t on their connector list, in different ways: Muse can build a custom connector for it or use its website in Muse’s own browser^[21][17], and a dot can sign into its website in the dot’s own browser or try your connected computer.^[47][11]

Meta’s Connectors help page says that if you want a service not yet on the Connector list, you can ask Muse to create a Custom Connector. Muse guides you through the setup, which can involve getting API information from the service, and stores these details in its Secure Credentials Store. Meta says it doesn’t review custom connectors or how they use your information, and asks you to grant access with caution and review the provider’s privacy policies.^[17] Meta’s business help page lists two more ways: set up a connection yourself if the service allows it, or let Muse use the website in its own browser, the way you would.^[21] The muse.ai site says that if a task needs a tool that doesn’t exist, the agent builds it.^[3]

OpenAI dots connect to apps through plugins.^[47] OpenAI’s safety post says you choose which accounts dots can use, either by connecting apps within the permissions you have granted to ChatGPT or by signing into a website in a dot’s browser, using secure sign-in where supported.^[11] OpenAI’s docs say that if a website blocks the cloud browser, you can ask your dot to try your connected computer. Trying your connected computer may create a separate local task, and it doesn’t transfer the cloud browser’s session.^[47]

To build your own plugin, OpenAI’s help center describes Plugin Creator, which brings skills, connected apps or both together in one plugin. Plugin Creator applies to all of ChatGPT, and the help page does not say a dot can create a plugin.^[50]

Yes, as of October 2026 Meta Muse and OpenAI dots can both write and run code: Muse writes its own code with a file system and a terminal and builds the tools a task needs^[19], and OpenAI dots run code and tools on their cloud computer and can start coding tasks in Codex.^[7][11]

Meta’s design post says: “Muse has its own computer, with a file system and a terminal, so it can write its own code and build the tools a task needs.”^[19] Meta’s security post says each Muse VM has enough storage, CPU and memory to do real work, like compiling code the agent writes and developing custom skills.^[16] Meta’s help page lists code files, “Scripts and source files in most programming languages”, among the file types Muse supports, and says Muse can build interactive single-page tools and visualizations you can open in your browser.^[51]

OpenAI’s safety post says each dot can create files and run tools on its cloud computer, and that sandboxing restricts what code and tools that dot can access. Dots run code apart from the systems that enforce safeguards, and “they cannot use that access to change those safety systems or turn off required checks”.^[11] OpenAI’s help center says your dot can create cloud tasks in Codex cloud environments.^[7] OpenAI’s docs say a new cloud coding task runs in a Codex cloud environment you have already set up, and your computer can be offline. A new local Work or Codex task runs on a computer connected to your dot, which must stay online with the ChatGPT app open.^[52] The docs also name GitHub as a plugin a dot can use to investigate an issue and prepare a pull request.^[47]

As of October 2026, Meta Muse, Meta's AI agent, and OpenAI dots, OpenAI's always-on agents in ChatGPT, can both make online purchases for you, and both need your approval for a purchase; Meta says Muse asks every time^[53], while OpenAI lets you give that approval in advance^[10] when it specifically covers the purchase.

Meta's help page on payments says Muse "will always ask for your approval prior to completing a purchase".^[53] One way Muse pays is by logging in to your account on a store's website with credentials kept in Muse's Secure Credential Store and using a card saved there, again after your approval.^[53] Meta's security overview adds that on a site where your card is already on file, Muse detects the checkout page and asks for approval "with the exact details of the purchase every time".^[16] Meta says purchasing ability may vary by location, merchant and checkout flow, and every order is fulfilled by the merchant.^[53]

OpenAI's dots privacy FAQ says a dot can make purchases using a card you've saved on a merchant's website, and that such purchases require your approval, "which may be given in advance when it specifically covers the purchase".^[10] OpenAI's getting-started article says you can set rules for what your dot can share, purchase or access.^[7] OpenAI's safety post says moving money between financial accounts is a step a dot must hand back to you.^[11]

Both vendors put the transactions on you. Meta's payments page says you're responsible for all transactions your Muse makes on your behalf.^[53] OpenAI's Terms of Use, which apply to all of ChatGPT for individuals, say you are responsible for all activities that occur under your account.^[29]

The practical difference: Meta Muse asks before every purchase it completes, and an OpenAI dot can act on an advance approval that specifically covers the purchase.

Meta Muse, Meta's AI agent, can pay with Link by Stripe, which Meta's help center^[53] lists as the recommended way to pay, and Link issues a one-time card number for each purchase so your real card number stays private, as of October 2026.

Meta's help page also says that when you use Link, Stripe offers coverage on eligible purchases "for damages, theft, loss, price drops, and returns".^[53] The muse.ai site says the one-time card number is generated at checkout, so your real card "stays hidden from the merchant and your agent".^[3]

Meta's security overview gives more detail. When Meta Muse pays from its wallet, a single-use card number is passed to the merchant's website instead of your regular card. Meta says the single-use card number is tied to one merchant, one dollar amount and a limited period of time, and Meta asks for your approval for every one of these payments.^[16]

Link is not the only way Muse pays. Meta's help page lists a second way: Muse logs in to your account with credentials from its Secure Credential Store and pays with a card saved on the site, after asking for your approval.^[53] Meta's launch post says Shop Pay is "coming soon" as another way to pay^[2]; as of October 2026 it is announced, not live.

Whichever way Muse pays, Meta's payments page says you're responsible for all transactions your Muse makes on your behalf, and suggests watching for email confirmations, receipts and statements.^[53]

As of October 2026, Meta Muse, Meta's AI agent, asks before sensitive actions like sending an email or making a purchase^[2], and OpenAI dots, OpenAI's always-on agents in ChatGPT, ask before actions like permanently deleting data^[11] and hand the most sensitive steps, such as changing a password or transferring money^[10], back to you.

For Meta Muse, Meta's help center splits actions into read actions, which view your information, and write actions, which make changes such as sending a message or making a purchase.^[43] With a connector set to "Ask for some actions", Muse asks before every write action and important read actions; with "Always ask", Muse asks before any action.^[43] On the web, "Ask for some actions" means Muse asks when your information may be shared or when Muse visits an unfamiliar website.^[43] Meta's security overview says read-only, previously allowed or low-risk actions can go ahead without asking^[16], and Meta's design post says the default allows standard web browsing and "stops for anything hard to undo".^[19] Meta adds that many approval checks are enforced outside the AI model.^[48]

For OpenAI dots, OpenAI's launch post says dots start with built-in rules for when to act on their own and when to ask for approval.^[5] OpenAI's privacy FAQ lists three levels: the most sensitive actions, like changing a password or transferring money, need you to take over and finish them yourself; actions like permanently deleting data or installing software may need approval each time; and some, like sending recurring messages, can be approved in advance.^[10] OpenAI's safety post says a dot asks for confirmation each time before permanently deleting data, installing or running software from an unrecognized source, or granting new security-sensitive access.^[11] The dots docs add that asking a dot to draft replies doesn't give the dot permission to send them.^[45]

Both products let you adjust part of this. In Muse you change when Muse asks for approval in Settings.^[17] A dot uses Custom Rules, which can't turn off core safety requirements.^[10]

In Meta Muse, Meta's AI agent, you change when Muse asks for approval in Settings, under Permissions^[43], where you set separate rules for connectors and for websites, as of October 2026.

Meta's help center says you can change when Muse asks for your approval in your Settings^[17]. Meta's page on guidance and approval lists three areas under Permissions^[43]:

• Connectors: "Ask for some actions" makes Muse ask before every write action and important read actions; "Always ask" makes Muse ask before any action.^[43]
• Web access defaults: "Ask for some actions" makes Muse ask when your information may be shared or Muse visits an unfamiliar website; "Always ask" makes Muse ask before accessing any website.^[43]
• Manage permissions: review permissions for each connector, manage approvals for each scheduled task and artifact, and see which websites you've allowed Muse to use and revoke them.^[43]

When Meta Muse asks in the moment, the prompt can offer Allow once, Allow for this task, Allow for this site, Always allow or Deny.^[43] "Always allow" lets Muse take that type of action for that connector in the future without asking again.^[43] Meta's security overview says Muse supports one-time, session, task, time-bounded or permanent permission.^[16]

Meta's help center adds that many connectors can be set up so Muse can only retrieve data, not take actions such as sending a calendar invite.^[17] Meta's design post says the built-in defaults can be changed to be "more or less cautious"^[19], and Meta's help center says Muse learns over time which decisions need your sign-off.^[43]

Custom Rules in OpenAI dots, OpenAI's always-on agents in ChatGPT, are optional settings that tell your dot how to handle a specific action: take the action without asking, take it if pre-approved, ask first, or hand it off to you, as of October 2026.^[7]

OpenAI's getting-started article says rules can cover what your dot can share, purchase or access. On mobile, you open Customize → Custom rules, describe the action the rule covers, and choose one of four behaviors^[7]:

• Take action without asking
• Take action if pre-approved
• Ask before taking action
• Hand off to you

"Pre-approved" means you explicitly requested the action in your prompt.^[7] OpenAI's dots docs call the second choice "Take action when you say so" and place the rules under Settings > Personalization > Custom rules.^[45]

OpenAI's privacy FAQ says Custom Rules can't turn off core safety requirements, such as a dot asking you to take back over to change a password, and can't change Auto-review or the limits on proactive research.^[10] OpenAI's safety post says dots can help you write Custom Rules but need your approval to change them.^[11] The docs say rules are instructions a dot tries to follow and can get wrong; they don't grant access to an app or computer or remove required confirmations, such as approval to use a saved login.^[45]

In ChatGPT Enterprise workspaces, the "Use custom rules for dots" permission is off by default, and default action rules still apply when custom rules are unavailable.^[15] If a workspace disables custom rules, saved rules can't be edited and don't apply.^[45]

Sentinel is a separate agent that runs on the same cloud machine as Meta Muse, Meta's AI agent, and nothing Muse does reaches the internet unless Sentinel approves it, according to Meta's launch post^[2] as of October 2026.

Meta says Sentinel is kept apart from Muse at the system level and asks you for permission when needed.^[2] Meta's security overview calls Sentinel "the sole permission authority" for actions through connectors to third-party services and for all network traffic leaving the machine: Muse proposes actions, but only Sentinel can grant permission.^[16]

For a connector action, Sentinel checks the connector policy you have set and decides whether to allow the action, deny it or ask you. Every network request is also checked by Sentinel on its way out. Read-only, previously allowed or low-risk actions can go ahead without a prompt.^[16]

When Sentinel needs your answer, the approval dialog appears in the Muse app's interface, not in your chat with Muse, and your answer goes straight back to Sentinel.^[16] Meta describes these approvals as "strict capabilities, not conversational suggestions", tied to a particular connector or destination and use, and says they can be one-time, session, task, time-bounded or permanent.^[16]

Sentinel also handles passwords and tokens. Meta Muse never sees real tokens, only stand-in ("surrogate") ones, and Sentinel puts in the real credential at the network boundary after a request is approved.^[16] Meta's help center adds that many approval checks are enforced outside the AI model, so they don't rely only on Muse deciding whether it should ask.^[48]

Auto-review in OpenAI dots, OpenAI's always-on agents in ChatGPT, is a separate safety system^[11] that checks certain planned actions^[10], such as sending an email or changing files, against your instructions, Custom Rules and safety requirements before they run, as of October 2026.

OpenAI's dots docs call this step an "automatic review" and say it decides whether an action can proceed, needs your approval, or includes a step you must do yourself; for example, you must change a password yourself.^[45]

For an email, Auto-review checks the recipient and the message to help catch a wrong address or information you did not intend to share.^[10] If Auto-review allows a step, the dot can rely on approval you've already given when that approval covers the action and the rules don't require a new confirmation.^[11] If Auto-review blocks a step, the action doesn't run and the dot is told why.^[11] The dot may then ask for clarification or approval, try a permitted alternative, or stop, and your approval can't override core safety requirements.^[10]

Ordinary read-only steps don't need this extra review, though they still follow app permissions, tool restrictions and other safeguards.^[11] OpenAI keeps the controls that enforce Auto-review outside the environments OpenAI dots can change, so a dot can't change or turn off a required check.^[11] Custom Rules can't change Auto-review either.^[10]

As of October 2026, Meta Muse, Meta's AI agent, keeps an Activity log^[43] of the actions Muse has taken and the permissions you've given, and OpenAI dots, OpenAI's always-on agents in ChatGPT, show ongoing and delegated tasks in Activity View^[10] in the ChatGPT desktop app.

Meta's launch post says Muse shows a complete audit trail of everything Muse has done and plans to do.^[2] In the app, you open the Activity log by tapping your assistant icon.^[43] Meta's help center also lists live status updates on your assistant avatar, an Upcoming view for scheduled tasks and reminders, and System Files "for a deeper look into your Muse's activities".^[43] The Activity log shows which skills Muse used^[54], and Meta's business help page says the log lists the steps Muse took and the actions Muse completed.^[21] Meta's security overview says you can see what Muse is doing in its browser and take over at any time.^[16]

OpenAI's privacy FAQ says Activity View shows a dot's ongoing and delegated tasks and lets you review steps the dot has taken.^[10] OpenAI's getting-started article says the dot's profile on the desktop app sorts activity into In progress, Scheduled and Completed, and scheduled tasks can also be managed from the Scheduled section in ChatGPT.^[7] The dots docs say that when a dot is waiting for a decision, app connection, sign-in or approval, you open the request and respond.^[45] Under Computers in the dot's profile you can inspect its cloud computer; opening the computer doesn't give you control until you select Take over.^[47]

Meta Muse, Meta's AI agent, and OpenAI dots, OpenAI's always-on agents in ChatGPT, can each undo some actions they have already taken, but both vendors say some actions can't be reversed, as of October 2026.

Meta's help center says you can ask Muse in chat to undo or stop some actions, like reading a website or comparing prices for a purchase, while "some other actions, like sending an email, cannot be reversed, just like if you were doing them yourself".^[43] Meta's muse.ai/business page answers the question of whether Muse will post or spend without asking with "Nothing publishes, sends or spends without your approval, and you can undo it".^[32] On a Mac, Meta says a file Muse deletes goes to your Trash, where you can restore it.^[55]

OpenAI's privacy FAQ says whether a completed action can be reversed "depends on the action and the app involved", and that some actions cannot be undone.^[10] The FAQ says a dot may be able to reverse unintended edits to a document or recall an email.^[10] OpenAI's dots docs add that stopping work doesn't undo completed actions, and that deleting a dot doesn't undo changes already made in connected apps or recall messages already delivered to other people.^[45]

On email the two vendors' pages differ: Meta says sending an email with Muse cannot be reversed, while OpenAI says a dot may be able to recall an email.

Both vendors say their agents can make mistakes, and as of October 2026 both tell you to step in, correct the work and report the problem: Meta Muse, Meta's AI agent, has a Report an issue^[48] option in Settings, and OpenAI dots, OpenAI's always-on agents in ChatGPT, point you to OpenAI's support page for dots^[10].

Meta's security overview says "Muse can and will still make mistakes"^[16], and Meta's getting-started page says Muse may be inaccurate or take unexpected actions, so you should monitor its actions carefully and intervene as needed.^[33] Meta's help center tells you to correct mistakes promptly and says you're responsible for fixing errors, including with third-party services or recipients.^[43] To report a problem on the web, open Settings and select Report an issue; in the app, shake your phone to open the reporting sheet, or go to Settings and select Report an issue.^[48]

OpenAI's launch post says dots can still make mistakes, "so always review consequential work".^[5] OpenAI's privacy FAQ says you can add context, correct a misunderstanding, change direction or ask your dot to stop, and if you spot an error, you can ask your dot to fix the problem.^[10] The dots docs warn that a completed run doesn't by itself confirm that the requested result was achieved or delivered, so review the output and any reported errors.^[52] To report a mistake, the FAQ points to OpenAI's support page for dots.^[10]

Under both vendors' terms, you are, as of October 2026: Meta's Muse terms^[13] say you are solely responsible for the actions Meta Muse takes for you, and for OpenAI dots, OpenAI's Terms of Use^[29], which apply to all of ChatGPT for individuals, say you are responsible for all activities under your account.

Meta's Muse Supplemental Terms, last updated September 8, 2026, say you are solely responsible for the actions Muse takes or directs in following your instructions, goals and guidance, and for artifacts it creates, publishes or hosts at your direction. For payments and purchases, you "bear all risk of financial loss". The terms also say you will keep meaningful human oversight over Muse's actions^[13], and Meta's help center adds that you're responsible for fixing errors, including with third-party services or recipients.^[43]

OpenAI's Terms of Use, effective January 1, 2026, apply to ChatGPT and OpenAI's other services for individuals and don't mention dots by name. OpenAI's terms say you are responsible for all activities under your account and for Content, meaning your input and the output you receive, and that you must evaluate output for accuracy, "including using human review as appropriate", before using or sharing it.^[29] ChatGPT Business and Enterprise, where dots are also offered^[9], use a separate OpenAI Services Agreement, which says the business customer is responsible for all activities under its account, including its end users' activities.^[56]

No, not in their supported sign-in flows: as of October 2026, Meta says Meta Muse, Meta's AI agent, has no visibility into your passwords or payment methods^[2], and OpenAI says OpenAI dots, OpenAI's always-on agents in ChatGPT, can use saved passwords without exposing them to the model^[5].

Meta's launch post says credentials you share with Muse go into secure storage, so Muse can use them without seeing them, and Muse can't see passwords you type into the browser yourself.^[2] Meta's help center says logins sit in a Secure Credentials Store; before Muse uses them, checks confirm that Muse is connecting to the intended service, that you connected or authorized the service, and that you gave Muse permission for that type of action.^[48] Meta's security overview says the agent only sees stand-in ("surrogate") tokens, and the email connector filters out one-time codes, password reset links and login magic links.^[16]

OpenAI's safety post says saved-password flows run through a dedicated encrypted credential service.^[11] For supported sign-ins, a dot pauses while you type your credentials into a secure login form; OpenAI says these sign-in protections don't cover passwords you share separately in a chat, a document or a plugin.^[10] The dots docs say you can also take over the dot's browser and sign in yourself, saving a login to Passwords is optional, and reusing a saved login needs your confirmation.^[47] You may need to give an authentication code or take over to finish a security check.^[10]

Meta Muse, Meta's AI agent, and OpenAI dots, OpenAI's always-on agents in ChatGPT, both use several layers against prompt injection, meaning instructions hidden in a website, email or file that try to steer the agent, and as of October 2026 both vendors say the risk remains: Meta calls prompt injection an open problem^[16], and OpenAI says its protections reduce the risk but do not eliminate it^[10].

Meta's help center lists model training, automated detection systems, and limits on the information and services Muse can access during a task, and says important permission and security checks run separately from the AI model.^[48] Meta's security overview adds that Muse's browser sub-agent sees an accessibility-tree snapshot of each page rather than the raw page code and can't run JavaScript; Meta checks its list of known harmful sites inside your VM and stops the browser from going there; and the agent never sees real tokens, so trying to coax secrets out of it is "futile".^[16] Meta says prompt injection "remains an open problem in the industry" and Muse will sometimes make mistakes.^[16]

OpenAI's privacy FAQ says a dot is designed to tell your instructions apart from content it reads while working, and that content "does not grant permission on its own".^[10] Other protections include limits on what tools can do, automatic checks before certain actions, approval requirements and safety monitoring.^[10] OpenAI's safety post says dots run code apart from the systems that enforce key safeguards and can't turn off required checks.^[11] OpenAI says these protections reduce the risk of malicious instructions causing an unwanted action, "but they do not eliminate it".^[10]

As of October 2026, Meta Muse, Meta's AI agent, and OpenAI dots, OpenAI's always-on agents in ChatGPT, are both built to check with you before sharing your information with other people: Meta says Muse won't share your artifacts unless you approve^[51], and OpenAI says a dot checks that you've allowed it^[35] before sharing details from a private conversation.

Meta's help center says Muse will not share your artifacts with others unless you turn on sharing yourself or approve the request when Muse asks.^[51] With web access set to "Ask for some actions", Muse asks for permission when your information may be shared.^[43] Meta's business help page says Muse checks with you before you publish, send, spend or share, and tells business users to use Muse only with information and accounts they are allowed to access.^[21]

OpenAI's dots docs say a dot checks that you've allowed it before sharing information from a private conversation with other people^[35], and that being able to use information doesn't give a dot permission to disclose it to another audience, such as a team channel.^[52] OpenAI's privacy FAQ says approving one message doesn't give a dot ongoing permission to contact people for you.^[10] OpenAI's safety post adds that health data always needs a named recipient, while less sensitive data, like an email address or phone number, needs a class of recipients by default.^[11]

As of October 2026, Meta's training setting for Meta Muse, Meta's AI agent, is on when you first use Muse and you can turn it off^[57]; on personal ChatGPT plans that include OpenAI dots, OpenAI says your content may be used to train its models^[10], with an opt-out^[12]; and on ChatGPT Business, Enterprise and Edu workspaces, the dots help center says "we don't use your data to train our models by default"^[10].

The Muse Privacy Policy says you can change the setting at any time and that "changes to this setting also apply to previous interactions".^[57] The same toggle covers information from connectors.^[57] To turn the setting off, go to Settings, then Data controls, toggle off "Help improve our AI models" and select Turn off.^[18] If you allow training, Meta says it removes certain personal identifiers, such as names, email addresses, phone numbers and Social Security Numbers, and disassociates the interactions from your account.^[48]

For personal ChatGPT plans, OpenAI's dots privacy FAQ says the "Improve the model for everyone" setting controls whether your dots' conversations and work may be used, including actions dots take, work they delegate, automations you set up and data from connected apps; OpenAI says it removes personal identifiers where possible.^[10] OpenAI doesn't train directly on a dot's proactive background research, but research brought into an eligible conversation may be used, depending on your settings.^[10] You turn the setting off in Settings, then Data controls; under a rule that applies to all of ChatGPT, your new conversations won't be used for training once the setting is off.^[58] For ChatGPT Business, Enterprise and Edu workspaces, the dots FAQ says: "we don't use your data to train our models by default".^[10]

One difference in timing: Meta says turning the setting off also applies to past Muse interactions, while OpenAI's ChatGPT rule speaks of new conversations.

As of October 2026, Meta says Meta Muse, Meta's AI agent, doesn't share your conversations or virtual machine data with Meta's ad systems^[57], and OpenAI says it doesn't share ChatGPT conversations with advertisers and shows no ads on Pro, Business or Enterprise accounts^[59], the plans that include OpenAI dots.

The Muse Privacy Policy says your conversations and the data in your virtual machine aren't shared with Meta ad systems^[57], and Meta's help center says this applies even if your Accounts Center includes other Meta products.^[48] Meta's business help page says Meta does not use your Muse conversations, or the files it works on for you, to choose the ads you see.^[21] Meta also says the agent's actions for you can still affect your ads the way your own activity would: if Muse visits a website or posts for you, that can affect the ads you see^[21], and there are "some legitimate scenarios where how you use Muse can influence the ads you see".^[16]

OpenAI's ads help page, which applies to all of ChatGPT, says ads may appear on the Free and Go plans, while Plus, Pro, Business, Enterprise and Edu accounts will not have ads.^[59] OpenAI's dots page says dots are available on Pro, Business Premium and Enterprise plans.^[9] The ads page says OpenAI does not share your ChatGPT conversations with advertisers, never sells your data to advertisers, and gives advertisers no access to your chats, chat history, memories or personal details.^[59]

Yes, in limited cases: as of October 2026, Meta says the design of Meta Muse, Meta's AI agent, limits Meta staff access to your data by policy but doesn't prevent Meta from accessing it when needed to support, secure or run the service^[16], and OpenAI says people associated with OpenAI may review OpenAI dots activity in limited circumstances^[10], including safety cases, even with model improvement turned off.

Meta's security overview says today's Muse architecture "restricts access to your data by Meta personnel through operational policies" but "does not prevent Meta from accessing data when necessary to support, secure or operate the service".^[16] Meta's Muse terms say Meta may monitor, log, review, suspend, block or modify Muse's actions, and may access and keep logs of actions Muse takes.^[13] Meta has announced Muse Confidential VM, which it says is meant to cryptographically prevent Meta from accessing data in your VM; Meta says a small group of trusted testers is using it.^[16] As of October 2026, Confidential VM is announced, not live.^[2]

OpenAI's dots privacy FAQ says human review "may occur in limited circumstances, including safety-related cases, even when model improvement is turned off".^[10] OpenAI's consumer-data help article, which applies to all of ChatGPT, says a limited number of authorized staff and trusted service providers may access content only to investigate abuse or a security incident, give support you ask for, handle legal matters, or improve models unless you've opted out, and that OpenAI monitors and logs all access to user content.^[60] The dots FAQ says OpenAI uses service providers for hosting, cloud infrastructure, support, safety and email delivery.^[10]

As of October 2026, Meta Muse, Meta's AI agent, keeps your data in a dedicated cloud virtual machine (VM)^[16] that Meta calls the system of record for everything you put in Muse, and for OpenAI dots, OpenAI's always-on agents in ChatGPT, OpenAI's rule for all of ChatGPT^[60] is that content is stored on OpenAI's and its service providers' systems in the US and around the world.

Meta's security overview says Muse sends limited data out of your VM only when needed for inference and telemetry.^[16] OAuth tokens for services you connect are stored in your VM, not in central Meta infrastructure.^[16] Meta's help center says Muse keeps backups to protect your agent's history from data loss^[48], and the security overview says your VM data is backed up continuously so you can restore it if something goes wrong.^[16]

OpenAI's dots privacy FAQ says a dot keeps context from your conversations and plugins for as long as you keep the dot, but that context doesn't keep credentials, images or screenshots.^[10] Files a dot creates or uses in other places, such as ChatGPT Library, your computer or a connected storage service, follow that location's retention policies.^[10] For ChatGPT Enterprise, OpenAI's docs say that during the beta, dots don't support data residency or inference residency, and dots don't provide strict zero data retention.^[44] OpenAI's privacy policy, which applies to all of ChatGPT, says personal data you choose to delete is removed from OpenAI's systems within 30 days unless OpenAI needs to keep it longer for reasons the policy lists.^[61]

As of October 2026, you reset Meta Muse, Meta's AI agent, in Settings, then Data Controls, then Reset Muse^[18], and in OpenAI dots, OpenAI's always-on agents in ChatGPT, you reset your dot from the ••• menu in the dot's profile^[7]; both resets delete the agent's chats and memories and both vendors say the step can't be undone.

Meta's help center says resetting Muse returns Muse to its original state and permanently deletes all information, including chat history, files and active tasks, and that "Resetting Muse cannot be undone".^[18] The steps are Settings, Data Controls, Reset Muse, then Reset to confirm.^[18] Short of a reset, you can delete any message, side chats and files in your VM.^[48] Meta notes that after you delete something, Muse may still remember information it learned from what you deleted.^[18] Deleting your Meta account from Accounts Center deletes all information tied to it, including Muse.^[18]

For OpenAI dots, OpenAI's getting-started article says Reset deletes your dot, including its conversations, saved memories and scheduled tasks. You open the dot's profile, select the ••• menu, select Reset, review what will be deleted and confirm; you then return to a new ChatGPT chat and can create a new dot in the desktop app or on desktop web.^[7] OpenAI's dots docs call this step Delete, say it cannot be undone, and say it doesn't undo changes already made in connected apps or recall messages already delivered.^[45] Files, Codex threads and ChatGPT conversations the dot created are stored separately and aren't deleted, and memories from your other ChatGPT conversations stay in place.^[10]

As of October 2026, Meta Muse, Meta's AI agent, has a Download your Muse data^[18] option in Settings, while for OpenAI dots, OpenAI's always-on agents in ChatGPT, OpenAI's dots privacy FAQ points you to a privacy request^[10] through OpenAI's Privacy Portal or by email.

Meta's help center says you can download a copy of your chats, files and other agent information: go to Settings, select Data controls, tap Download your Muse data, then select Download to confirm.^[18] Meta's security overview says you can inspect, edit and download the files in your VM freely, including Muse's memory about you.^[16] For your Meta account information, the help center describes a separate export in Accounts Center, under Your information and permissions, then Export your information.^[18]

OpenAI's dots privacy FAQ says you can submit a request through OpenAI's Privacy Portal or email dsar@openai.com, even if you don't have an OpenAI account.^[10] Depending on where you live, your rights may include receiving a copy of your personal data.^[10] A separate rule applies to all of ChatGPT: signed-in Free, Go, Plus and Pro users can request a copy of their data from ChatGPT settings, while self-service export isn't available for ChatGPT Business or Enterprise workspaces, where requests go to the workspace owner.^[58] OpenAI's data-controls help page doesn't mention dots, so the page doesn't say whether a ChatGPT data export includes a dot's own context.

Yes: Meta Muse, Meta's AI agent, can be used with a separate account, and if you keep accounts separate, the Muse Privacy Policy^[57] says Muse won't share your Muse interactions or the data on your virtual machine with other Meta products, as of October 2026.

Meta's Muse Supplemental Terms say you can keep Muse in a separate Accounts Center by signing up with an email address that isn't tied to another Meta account.^[13] Meta's help center says the Facebook, Instagram and Threads connectors are connected to Muse automatically if your accounts are in the same Accounts Center.^[17]

If you do add Muse to the same Accounts Center as other Meta products, the Muse Privacy Policy says Meta may combine and use your information across profiles and devices in that Accounts Center, except where the policy says data will not be used.^[57] Meta's help center says Muse doesn't share your conversations or VM data with Meta ad systems, even if your Accounts Center includes other Meta products.^[48]

To remove everything, deleting your Meta account from Accounts Center deletes all information tied to that account, including Muse.^[18]

Yes, as of October 2026 Meta Muse and OpenAI dots each run on their own computer in the cloud, with their own browser, and you can open that computer to watch the work or take over.^[47][16]

Meta’s Muse runs on Muse Secure VM, a dedicated computer in the cloud that Meta says is contained so no one else’s agent can reach it^[2]. Meta’s site calls it a persistent, isolated Linux computer with a full browser that you and your agent can both use, so you can step in at any point^[3], with the storage, CPU and memory to handle real work^[3]. Muse’s browser is based on Chromium, and you can see what Muse is doing in it and take over at any time^[16]. Muse also has a file system and a terminal, so it can write its own code and build the tools a task needs^[19]. Meta says the VM is the system of record for everything you put in Muse, and Muse sends limited data out of it when necessary for inference and telemetry^[16]. For tasks that take more time, Muse keeps working after you close the app^[2].

OpenAI’s dots each have their own cloud computer and their own browser^[5]. OpenAI says it isolates users’ cloud environments from one another and maintains the Linux operating system and Chrome browser^[11]. On that computer a dot can browse, analyze information, create files and run tools, and sandboxing limits what code and tools it can reach^[11]. The dot’s computer keeps its state between uses and has its own files, software and browser sessions, so signing in to a website on your own computer doesn’t sign your dot in^[47]. To watch, open your dot’s profile and look under Computers; opening the computer doesn’t give you control, so you select Take over, then Return control^[47]. Your dot can keep working even when your own computer is off^[23].

Yes, as of October 2026 both can if you allow it: Meta Muse can work with files and some apps through the Muse app on a Mac, and OpenAI dots can work on one computer you connect through the ChatGPT app.^[47][55]

Meta’s help page says that with the Muse app on your Mac, Meta’s Muse can find, organize and manage the files you ask it to work with, and read information from apps like iMessage, Notes, Reminders, Email and Calendar^[55]. Muse asks for permissions that macOS sets and controls, not Meta: Full Disk Access, Automation and Notifications^[55]. Meta notes that Full Disk Access covers all the files on your Mac, but the agent only uses the files and app information you ask it to work with^[55]. For each app that keeps information on your Mac, you choose Off, Read only, or Read and interact^[55]. Muse asks you to confirm before important actions like sending an email or making a purchase, and files it deletes go to your Trash, where you can restore them^[55]. You can turn off this access in Muse’s Settings, or remove the permissions in your Mac’s System Settings^[55]. Meta says your interactions with Muse can include screenshots of your screen, and you decide whether those interactions are used to improve Meta’s AI models^[55]. Meta warns that computer use gives the agent broad reach across your Mac, so only grant access you’re comfortable with^[55].

OpenAI’s dots have their own cloud computer, and access to your local computer is optional and starts turned off^[7]. To connect, you open your dot’s profile in the ChatGPT app on that computer, find Your computer under Computers and select Allow access; you can connect only one personal computer at a time^[47]. Your dot can then access files and work on that computer from any of its messaging channels, and Revoke access stops it^[7]. The computer must be online with the ChatGPT app open for your dot to use it^[23]. If a website blocks the cloud browser, you can ask your dot to try your connected computer, which may create a separate local task^[47]. Camera, microphone or screen access needs a connected computer and device permissions for the ChatGPT app^[10]. OpenAI’s enterprise docs say conversations and task context don’t stay only on the connected computer^[44].

Yes, as of October 2026 Meta Muse and OpenAI dots both keep working in the cloud after you close the app or step away, and both come back to you when something needs your input.^[5][2]

Meta says that for tasks that take more time, Meta’s Muse keeps working after people close the app, and comes back when something changes or when it needs approval, like before it sends an email or makes a purchase^[2]. Meta’s business help page says Muse can keep multi-step or scheduled work moving after you close it^[21]. Meta’s design post says Muse can manage several tasks in the background while the app is closed, and a snippet under its avatar shows what it’s working on^[19]. When background work finishes, Muse decides whether the result is worth showing you and notifies you only when something is meaningfully new or needs your input^[19]. To stop something, you can ask Muse in chat to undo or stop some actions, like reading a website or comparing prices^[43].

OpenAI’s dots keep making progress between conversations^[7], and OpenAI’s docs say your dot lives in the cloud and can keep working even when your computer is off^[23]. OpenAI says dots can work toward your goals 24/7^[5]. Your dot can decide when to pause and wake up to continue work, and can use background agents to work on several things in parallel while you keep talking to it^[23]. Dots can also message you with progress, questions or decisions that need you^[5]. To stop your dot, you open the ••• menu in its profile and select Pause^[7]; OpenAI’s docs add that Pause stops your dot’s current main task but doesn’t stop every delegated task or cancel future scheduled runs^[45].

Yes, as of October 2026 Meta Muse and OpenAI dots can both set reminders and run recurring tasks on a schedule, and you set them up by asking in the chat.^[7][62]

Meta’s help page says that in Meta’s Muse you can set one-time reminders for a date and time, or location-based reminders for when you’re near a place, and create recurring tasks that repeat daily, weekly or on a custom interval^[62]. Muse reminders arrive as messages in your Muse conversation, times follow your time zone, and recurring tasks continue until you cancel them^[62]. To change or remove one, you tell Muse which reminder or task to update, or open the Upcoming tab under the Assistant icon to see all of them^[62]. Meta says that once you set a goal or a task, Muse also works on a schedule and in response to relevant events^[19]. In Muse’s Settings, under Permissions, you can manage your approvals for each scheduled task^[43].

OpenAI’s help page says you can ask your dot to set a reminder or run a recurring check^[7]. To manage a scheduled task in OpenAI’s dots, you select it in Recent activity in your dot’s profile or open the Scheduled section in ChatGPT, where you can change its repeat schedule, time or completion notifications^[7]. OpenAI’s docs say work that repeats at fixed times needs a saved schedule: what to check, when to run, including the time zone and any end date, which changes deserve a notification, and where to deliver results^[52]. Where a connected service supports event monitoring, you can ask your dot to respond to a specific event, but connecting a source alone doesn’t create a monitoring task^[52]. Your dot’s recurring tasks can include actions you’ve authorized, and their permissions are separate from proactive research^[52].

Yes, as of October 2026 Meta Muse and OpenAI dots both look for ways to help without a new request: Muse sends proactive messages and ideas, and a dot runs what OpenAI calls proactive research.^[5][19]

Meta’s design post says Meta’s Muse can send you messages without you prompting it^[19], and that you can tell your Muse to turn this off, dial it down, or dial it up^[19]. After background work, Muse notifies you only when something is meaningfully new or needs your input^[19]. The Ideas tab gives personalized task suggestions^[54]. For Muse for Small Business, Meta says Muse can come to you with ideas, like flagging emails that need a response and writing first drafts^[4]. Some connectors can share information with Muse on their own, and Meta says it identifies those connectors before you connect them^[48].

OpenAI says that when you aren’t actively working with your dot, it looks for ways to help in the background, which OpenAI calls proactive research^[5]. Proactive research uses your connected apps with read-only tools that can’t send messages, change app content, or control your browser or computer^[5], and OpenAI says it enforces these limits in code^[11]. Your dot can review connected information and form memories from it, even when you haven’t asked a new question^[7], and can save private notes; the same action rules apply to tasks you’ve authorized that continue in the background^[10]. OpenAI’s dots can also message you with progress, questions or decisions that need you^[5]. OpenAI’s help page says you can review scheduled work in your dot’s profile or pause your dot when you don’t want it to continue^[7].

As of October 2026, Meta Muse and OpenAI dots both remember things about you: Muse remembers what you share, what it learns about your preferences and what is in its memory files, and a dot remembers what it gets from your ChatGPT memory plus its own notes about you.^[52][39]

According to Meta, Meta’s Muse remembers what matters to a person, so it can make suggestions unprompted and act on details mentioned only once^[2]. Meta’s help page says Muse builds its understanding of you from what you share in conversations, preferences and patterns it observes, information from connectors you’ve connected, and its file-based memory^[39]. Muse’s memory files include a Memory file for information you share, a Soul file for who your Muse is, and an Identity file for details like its name, all under the Assistant icon, then Identity. You can ask Muse what it has saved about you^[18].

OpenAI’s help page says your dot receives memories from ChatGPT and can create its own memories, including from connected apps^[7]. Each of OpenAI’s dots starts with your ChatGPT memory^[9]. OpenAI’s docs say your dot also keeps its own notes about your preferences, decisions and ongoing work, separate from ChatGPT’s saved memory and not a complete transcript of what you’ve said^[52]. Your dot can keep what it learns from your conversations and plugins for as long as you keep it, but it doesn’t keep credentials, images or screenshots^[10].

As of October 2026, Meta Muse lets you see and edit what it remembers in its Memory.md file^[18], while OpenAI dots don’t currently let you see or edit individual dot memories; what a dot shares with ChatGPT Memory is managed in ChatGPT Memory settings.^[10]

Meta’s help page says you can edit and delete information in the Memory.md file directly, under the Assistant icon, then Identity, then Memory. Muse’s Soul.md and Identity.md files, also under the Assistant icon, then Identity, can be edited or cleared too. You can ask Muse what it has saved about you^[18], and you can ask Muse in the conversation to change what it remembers, or edit that information yourself^[39]. The forget skill asks Muse to look for information about a person, topic or matter in its memories and files and remove it as well as it can^[48]. Meta notes that after you delete something, Muse may still remember information it learned from it^[18]. Meta’s security post says you can inspect, edit and download the files in your VM, including Muse’s memory about you^[16].

OpenAI’s help page says you currently can’t view, delete or directly change individual dot memories, including details that enter its context from plugins; you delete your dot’s context by deleting your dot. Anything your dot has shared with ChatGPT Memory is managed separately in ChatGPT Memory settings^[10]. On mobile, Customize → Memory shows what ChatGPT remembers, using shared ChatGPT memory settings^[7]. Turning off Memory in ChatGPT stops memory sharing with OpenAI’s dots, but doesn’t delete what your dot already received^[10].

As of October 2026, OpenAI dots, OpenAI’s agents in ChatGPT, start with your ChatGPT memory, but each dot also keeps its own notes, which OpenAI says are separate from ChatGPT’s saved memory.^[52][9]

OpenAI’s help page says your dot receives memories from ChatGPT and can create its own memories, including from connected apps^[7]. OpenAI’s docs say your dot draws on the conversation, relevant ChatGPT memory and its own saved notes about your preferences, decisions and ongoing work^[23]. Your dot’s notes are not a complete transcript of everything you’ve said^[52].

Memory sharing between ChatGPT and OpenAI’s dots runs both ways: your dot can receive memories and recent conversation context from ChatGPT, and conversations with your dot can add to memory in ChatGPT^[10]. Turning off Memory in ChatGPT stops that sharing, but it doesn’t delete information your dot has already received^[10]. OpenAI’s docs add that changing a ChatGPT saved-memory setting doesn’t necessarily change the notes your dot has already made^[52].

ChatGPT memory and your dot’s memory also have different controls. Information your dot has shared with ChatGPT Memory is managed in ChatGPT Memory settings^[10], and on mobile, Customize → Memory shows what ChatGPT remembers^[7]. Your dot’s own memories can’t currently be viewed, deleted or changed one by one; you delete your dot’s context by deleting your dot^[10]. For ChatGPT memory in general, which applies to all of ChatGPT, turning off Memory doesn’t delete your past chats, and deleting a chat doesn’t necessarily delete a saved memory created from it^[63]. Your dot can retain context from your conversations and plugins for as long as you keep your dot, and that context doesn’t retain credentials, images or screenshots^[10].

As of October 2026, Meta Muse is powered by Muse Spark, a Meta model^[2], and OpenAI dots are powered by GPT-6 Astra, an OpenAI model.^[5]

Meta announced Muse Spark, the model behind Meta’s Muse, on April 8, 2026, as the first in a new series of large language models built by Meta Superintelligence Labs^[64]. Meta’s April post says Muse Spark also powers the Meta AI assistant in the Meta AI app and on meta.ai^[64]. Meta’s September 2, 2026 research post says the newest Muse Spark release is available to developers in Muse Code and Meta Model API^[65]. Meta says the newest Muse Spark release asks clarifying questions when a prompt is unclear and confirms before taking consequential actions^[65].

OpenAI introduced GPT-6 Astra, the model behind OpenAI’s dots, on September 3, 2026^[20]. OpenAI’s GPT-6 Astra post said the model would become available to all ChatGPT Plus, Pro, Business and Enterprise users over the following days, and through the OpenAI API, Microsoft Azure and AWS Bedrock^[66]. OpenAI says it trains GPT-6 Astra, the model that powers dots, to understand your intent and follow safety rules^[11], and OpenAI’s dots help page says GPT-6 Astra is trained to refuse harmful requests, including biological or cybersecurity misuse^[10].

As of October 2026, Muse skills are built-in capabilities, developed by Meta, that let Meta Muse do specific tasks such as creating podcasts, generating images, compiling research, drafting documents, creating dashboards or browsing the web.^[54]

Meta’s help page says skills are already part of Muse and are automatically available, so you don’t need to install them separately^[54]. When you give Muse a request, it works out which skills the task needs^[54]. For example, asking for a podcast about a topic uses the audio generation and web browsing skills, and asking for a PDF uses the document creation skill^[54]. Skills activate when you make a request that requires them^[54].

Some skills need information to work. Location-based skills may need your location, for example for a reminder to buy eggs when you’re next at the store^[54]; file-based skills need the files you upload or store in Muse; and web browsing skills go online to find information for you^[54]. For actions like sending emails or making purchases, Muse is designed to ask for your permission before going ahead^[54].

To find skills, Meta suggests the Ideas tab for personalized task suggestions, or asking your Muse “What can you help me with?”^[54]. You can review which skills Muse has used in your Activity log, under the Assistant icon^[54]. Meta’s security post says Meta wrote skills, which it describes as detailed instructions to Muse, for getting the most out of each connector Meta built^[16]. Meta’s Muse for Small Business announcement added a collection of new skills and connectors inside Muse to help people run their businesses^[4].

As of October 2026, Meta Muse works with documents, spreadsheets, images, code files and web files from a published list of supported types, and OpenAI dots work with files and photos you attach, files on the dot’s cloud computer and files in connected apps or on a connected computer, without a published list of file types.^[23][7][51]

Meta’s help page lists the file types Meta’s Muse supports: PDF, DOCX, plain text and Markdown documents; XLSX and CSV spreadsheets; PNG, JPEG, GIF and WebP images; and code files in most programming languages^[51]. Muse can also build interactive single-page tools and visualizations you open in your browser^[51]. Muse can read, summarize, edit or generate artifacts, including PDFs, spreadsheets, documents, images and podcasts, and finished artifacts are saved in the Library tab^[33], where you can delete files at any time^[51]. With the Muse app on a Mac, Muse can also find, organize and manage the files you ask it to work with^[55]. Muse won’t share your artifacts with others unless you turn on sharing yourself or approve its request^[51].

OpenAI’s help page says you attach a file or photo to a dot conversation by selecting +^[7]. OpenAI’s docs say your dot can research, analyze data, prepare documents and build software^[23], and OpenAI says each dot’s cloud computer can create files and run tools^[11]. If you connect your computer, its local files, code and apps become available for your dot’s work^[23], and with the Google Drive plugin, your dot can read and update documents^[47]. OpenAI’s dots pages don’t list supported file types. To share a file, OpenAI says dots are taught to seek authorization that covers the information and the type of recipient^[11]. Files your dot creates or uses elsewhere, such as ChatGPT Library, your computer or a connected storage service, follow that location’s retention policies, and deleting your dot doesn’t delete files it created^[10].

As of October 2026, you start Meta Muse by signing in with a Meta account in the Muse app or on the web^[33], and you start OpenAI dots by creating your dot in the ChatGPT desktop app or in ChatGPT on desktop web, on a plan that includes dots.^[7]

Meta’s help page says you connect to Meta’s Muse with your Meta account, and you can create one with a Facebook or Instagram account, an email address or a phone number^[33]. The Muse app is on macOS, iOS and Android, and the web version is at muse.ai^[21]. Muse is free to start, with a usage limit^[1]. You open Muse on the Chat tab and type a message or question^[33]; Muse isn’t turn-by-turn, so you don’t have to wait for a reply before your next question^[33]. When you describe a task, Muse may ask clarifying questions, then get to work^[33]. Meta notes that Muse may be inaccurate or take unexpected actions, so you should monitor it and step in as needed^[33].

OpenAI says your first dot is included in a Pro or Business Premium plan at no extra cost^[5], Pro 100 costs $100 a month^[6], and dots aren’t on ChatGPT Free, Go or Plus^[12]. You create your dot on desktop and follow the onboarding prompts; you can’t currently create a dot on mobile^[7]. OpenAI’s docs list the setup steps: open dots in ChatGPT and follow the introduction, connect apps such as email, calendar and files or skip that step, and in the desktop app choose whether to connect your computer^[38]. Your dot then introduces itself and suggests ways it can help^[23]. After setup, you can talk to it in the ChatGPT mobile app when mobile access is available^[7]. OpenAI’s dots are rolling out gradually, and access may take several days to reach your account^[7].

As of October 2026, the terms for Meta Muse and for OpenAI dots both bar making important decisions about people without human review, passing AI output off as human, and deceiving or impersonating others, and both say not to rely on the agent for professional advice.^[67][29][13]

Meta’s Muse Supplemental Terms say you may not use Muse to bypass a third party’s paywalls or CAPTCHAs, decide without meaningful human review whether someone gets or keeps a job, credit, housing, insurance or other important benefits, or start bulk calls or messages without required consent^[13]. The terms also bar misrepresenting Muse’s responses as human-generated, hiding that the agent acts for you where the law requires disclosure, and impersonating or deceiving any person or company^[13]. The terms say you will keep meaningful human oversight over Muse’s actions, and that you should not let it send messages, make payments or delete data without review and approval or pre-authorization^[13]. Meta says Muse doesn’t provide legal, financial, investment, medical or tax advice^[13].

OpenAI’s Terms of Use apply to all of ChatGPT, which is where OpenAI’s dots run^[7], and say you may not represent that output was human-generated when it wasn’t, or automatically extract data or output^[29]. The same terms say you must not use output about a person for decisions with a legal or material impact on them, such as credit, employment, housing, insurance or medical decisions^[29], and you should not rely on output as a substitute for professional advice^[29]. OpenAI’s usage policies, which also apply to all of ChatGPT, bar deceit, fraud, scams, spam or impersonation, and automation of high-stakes decisions in sensitive areas without human review^[67]. For ChatGPT Pro, the personal plan that includes dots^[12], OpenAI’s help page says sharing account credentials, reselling access and automatically extracting data are prohibited^[6].

As of October 2026, Meta and OpenAI have each announced plans that aren’t live yet: Meta has named AI glasses support, a Muse device, an email address and a private VM for Meta Muse, and OpenAI has said you’ll be able to add more OpenAI dots and is previewing specialist dots for businesses.^[7][46]

Announced for Meta’s Muse, not live yet:

• Muse on Meta’s AI glasses “in the coming months”, where you’ll be able to say your agent’s name and Muse can act on what you’re looking at^[46].
• Muse Charm, which Meta calls the first device made for Muse, with a sign-up to hear when it’s here^[68].
• An email address of its own that Muse can use to get things done^[46].
• Muse Confidential VM, which Meta plans to deliver in 2026 (“later this year” in Meta’s words) and which is intended to cryptographically and verifiably prevent Meta from accessing data in your VM; Meta says a small group of trusted testers already uses it^[16].
• Bigger models and a Muse Spark open-weights release, on Meta’s model roadmap^[65].
• Meta Enterprise Platform, announced September 28, 2026, through which Meta says it will bring “the Muse agent” to businesses, and that Chirantan “CJ” Desai will join Meta to lead it as Chief Enterprise Platform Officer^[69]. Meta gives no product, price or date for this.

Announced for OpenAI’s dots, not live yet:

• Adding more dots, and scaling each dot’s output by its speed or the amount of work it can take on per month^[7].
• Teams of dots working together, which OpenAI describes as its vision over time^[5].
• Expanding dots to more users soon^[5].
• Calls started by your dot, which OpenAI’s docs say are planned for after launch^[35].
• Specialist dots, shared as a preview, with their own identity, IT-provisioned hardware and deep integrations with a company’s systems of record, starting with focused enterprise pilots^[5].
• Work with Microsoft to bring specialist dots into Agent 365, with the goal of letting businesses manage dots through the Microsoft tools they already use^[5].