How to turn off Meta AI, and what you can do instead

The short answer

As of October 22, 2026, we did not find a Meta help article that describes a setting to turn off or remove Meta AI in Facebook, Instagram, WhatsApp or Messenger. WhatsApp's help center lists Meta AI among the optional services from Meta.

What Meta's pages do describe: deleting your chats with Meta AI, typing commands that delete Meta AI's copy of your messages, turning off message sharing with Meta AI in individual Messenger chats, turning off some WhatsApp AI features, and, in the EU and UK, objecting to the use of your information for AI training.

Where Meta AI shows up

Facebook's help center says you can reach Meta AI from your Facebook messages by tapping Messenger, then Meta AI. Instagram's help center lists three ways in: a conversation with Meta AI, Meta AI in chats with others, and Meta AI in search. WhatsApp's help center says people can mention @Meta AI in group chats.

Meta's help pages say these features are being introduced gradually and may not be available to everyone.

Delete Meta AI's copy of your messages (Messenger, Instagram, WhatsApp)

Meta's Privacy Center says two commands reset its AIs by deleting their copy of your messages and any saved details about you. You still see your own copy of the chat, but the AI won't remember earlier messages. After using a command, you can delete the chat from your view with that app's normal delete steps.

WhatsApp's help center adds that deleting a chat in WhatsApp deletes it only on WhatsApp; to delete your messages from Meta, you need the commands. It says the commands don't work in Incognito chats with Meta AI, because those messages are not saved.

• To reset one AI: open your individual chat with the AI and type /reset-ai.
• To reset the AIs across that app, including AIs in group chats you're in: type /reset-all-ais in an individual chat with an AI.

Steps by app

These are the steps in Meta's help centers as of October 22, 2026. Menu names can differ by device and app version.

• WhatsApp (Android or iPhone), delete a chat with Meta AI: open the chat with Meta AI, tap the chat history icon (a clock with a circular arrow), tap and hold the chat you want to delete, then tap Delete chat, then Delete. WhatsApp says a deleted Meta AI chat can't be restored.
• WhatsApp, reset Meta AI: open the chat with Meta AI, tap the chat history icon, tap the chat, then type /reset-ai in the message field and send it. WhatsApp says /reset-all-ais also resets your other individual and group chats with Meta AI.
• Messenger (iPhone), delete a chat: from Chats, swipe left on the conversation, tap More, tap Delete, then tap Delete to confirm. Messenger's help center says deleting is permanent; to hide a chat without deleting it, archive it instead.
• Messenger (computer), turn off message sharing with Meta AI for a chat: open the chat, click Information, click Message permissions under Privacy & support, then turn off Allow message sharing. In a group chat, only admins can change this.
• Instagram (iPhone), delete a chat: tap Messages, tap and hold the conversation, tap Delete, then Delete again. Instagram says this deletes the chat only for you.
• Facebook (computer), hide a chat: click Messenger at the top right, hover over the chat, click Options, then Archive chat. Facebook's help center says an archived chat comes back to your inbox if the person messages you again.
• Meta AI app, delete chats and media: tap Menu, then Settings, then Data & privacy under App settings, then Manage your information, then Delete all chats and media, then Delete all. Meta says this does not delete public vibes you posted.

Limit what Meta AI can read

WhatsApp's help center says that in group chats, Meta AI can read only messages that mention @Meta AI, and that your personal messages and calls remain end-to-end encrypted. It says Private Processing features, which power message summaries and writing help, are off by default. On Android, it says to go to Settings, then Chats, then Private Processing to turn them on or off.

WhatsApp also offers Incognito chats with Meta AI. Its help center says only you can see these chats, Meta and WhatsApp can't access them, and by default the conversation disappears when you leave.

On Messenger, Meta's help center says that when people use AI features in a chat, up to 50 recent messages are shared with Meta AI, or messages from the past 30 days in some cases, such as mentioning @Meta AI in a group chat. Turning off message sharing permissions stops people in that chat from using those AI features, though Meta notes people can still forward messages to Meta AI.

On Instagram, Meta's help center says messages from the past 30 days are shared when people use AI features in your chats. It says that if you don't want others to share certain messages with Meta AI, you can delete the chat or the messages.

Can you stop Meta from training AI on your data?

Meta's Privacy Center says it trains its AI on information shared on its products, including your interactions with its AI features. It says it does not use your private messages with friends and family for training unless you or someone in the chat shares them with its AIs.

• European Union: on April 14, 2025, Meta said EU users would get in-app and email notices with a link to a form to object at any time, and that it would honor objections already received. It said it does not train on public data from EU accounts of users under 18.
• United Kingdom: on September 13, 2024, Meta said adult UK users on Facebook and Instagram would get notices with a way to reach an objection form, and that it would honor existing and new objections.
• United States: Messenger's help center links to a form to submit an objection, which requires logging in to Facebook. When we opened Instagram's help article on objections from the U.S. on October 22, 2026, it said the information is not available in our area. The Privacy Center page, viewed from the U.S., offers a form for requests about personal information from third parties used for AI. We did not find a Meta page that describes a general way for U.S. users to opt out of AI training on their own activity.

How Meta uses your AI chats for ads and content

Meta announced on October 1, 2025 that, starting December 16, 2025, it would use people's text and voice interactions with its AI features to personalize the posts, reels and ads they see, in most regions. Meta said conversations about topics such as religious views, sexual orientation, political views or health would not be used to show ads. The announcement does not describe a way to opt out; it points to Ads Preferences and feed controls.

Meta said that if a WhatsApp account is not added to the same Accounts Center as other Meta accounts, Meta AI interactions on WhatsApp won't personalize experiences on those accounts.

Meta's help center also says Meta uses interactions with its AIs to improve AI at Meta, and that if the AIs can't answer a question, Meta shares the messages and general information such as your region with select partners.

Teens and parents

On January 23, 2026, Meta said it was pausing teens' access to existing AI characters globally while it builds a new version. Teens can still use the Meta AI assistant, which Meta says has default age-appropriate protections.

Meta said on April 23, 2026 that parents who supervise a teen account on Facebook, Messenger or Instagram can see the topics their teen asked Meta AI about over the past week, but not the conversations. On October 5, 2026, Meta said supervision is available globally in the Meta AI app, where parents can switch a teen from the default 13+ setting to a Limited Content setting. For more, see our guide to whether Meta AI is safe.

Government inquiries and laws

These are inquiries and allegations, not findings.

• FTC: on September 11, 2025, the FTC sent study orders under its 6(b) authority to seven companies, including Meta Platforms and Instagram, asking among other things how they use or share personal information from user conversations. It is a study, not an enforcement case.
• Texas: on August 18, 2025, the Texas attorney general announced an investigation of Meta AI Studio and Character.AI over potentially deceptive trade practices. The release claims, among other things, that the platforms' terms of service show user interactions are logged and used for targeted advertising.
• 44 attorneys general wrote to 13 AI companies, including Meta, about sexualized AI interactions with children (August 25, 2025).
• California SB 243, in effect since January 1, 2026, requires companion chatbot operators to keep a protocol that refers users who express suicidal ideation or self-harm to crisis services. New York's AI companion law requires operators to tell users at the start of an interaction that they are not talking with a human. Whether Meta AI counts as a companion chatbot under either law is a legal question.

We are not saying Meta is covered by, or out of compliance with, any of these laws. It may have practices it has not published, and whether a law applies to it is a legal question this guide does not decide.