Jump to a section
Choose the job first. Avoiding an AI chat, switching off a photo suggestion and objecting to the use of public posts for training are different actions. UK adults can use Meta's stated objection route; do not treat that as proof that every AI feature has gone or earlier training has been undone.
What do you want to stop?
Start with the particular thing you do not want. A blue-circle assistant, an AI search answer and a suggestion to edit a photo are different features. Closing one does not establish a preference for everything using AI.
This guide brings the shared Instagram and Facebook choices together. Use the table to choose a next step rather than spending time looking for one account-wide off switch.
| Your goal | What to do |
|---|---|
| I do not want to chat with Meta AI | Leave the assistant unused and check who or what you are messaging before sending. |
| I do not want my public posts used for AI training | Read Meta's UK notice and use its objection route. This is separate from an AI chat. |
| I do not want suggested edits from my camera roll | On Facebook, review that feature's own control if it is available in your account. |
| I want to remove something I already posted | Review the post and its audience separately. A training objection is not a post-deletion request. |
The UK training objection applies to both apps
Meta's UK notice, originally published in September 2024 and checked for this guide on 12 November 2026, describes using public information from adult Facebook and Instagram accounts for generative-AI training. It includes public posts, comments, photos and captions.
The notice says people in the UK can object at any time and that Meta honours existing objections. It describes a data-use choice, not a switch that removes the AI interface. Read the current notice and form for your account before submitting.
Use the official route, then check what it covers
Meta's official Generative AI Privacy Guide is linked below. It may ask you to sign in. You can also start with the AI training notice shown in your own app. Use those official routes rather than an unsolicited message claiming there is an urgent opt-out deadline.
The exact signed-in form and menu position were not available for this source check, so these are preparation steps rather than a claimed tap-by-tap walkthrough.
- Open the official AI privacy guide or the AI training notice within the app you use.
- Find the objection information for the UK and read the data and accounts it covers.
- Submit the objection if that is your choice. Save any confirmation you receive and its date.
- If you use more than one Facebook or Instagram account, check the form's coverage for each one. Do not assume a submission covers every profile.
- Keep using the app normally if you wish, while treating any new information you send to an assistant as a separate sharing choice.
On Instagram: review the public content too
For Instagram, the public photos, captions and comments described in Meta's notice are the relevant starting point. That is a different question from whether you have ever opened the assistant or used an AI editing tool.
Review the material you have made public and decide whether you still want it visible. If you change an audience or remove an old post, treat that as a separate privacy action from submitting the training objection.
Before sending a message, check the recipient and whether you are in an assistant conversation. You do not need to send a private photo or copy a friend's messages into AI to keep using ordinary Instagram features.
On Facebook: photo suggestions are a separate choice
Facebook has more than one AI feature. Meta's 15 June 2026 announcement describes an AI Mode in search as well as AI-assisted camera-roll sharing suggestions. It says the camera-roll suggestions are opt-in and can be switched off.
If your account offers that photo feature, use its own camera-roll control and the linked Facebook help. This changes that feature; it is not an objection to the use of public posts for training.
Availability can differ by account and country. The announcement does not establish that every UK account has the feature, and we have not observed the signed-in menu on your phone. Do not turn on a feature you have never used just to look for its off control.
Keep the different choices separate
Deleting a conversation, hiding a notification, removing a post and objecting to training are different requests. Do not take one successful change as proof that the others have happened.
The practical privacy habit is the same after making an objection: keep passwords, security codes, bank details and other people's private information out of prompts. A data-use preference is not a reason to share more than the job needs.
If your goal is to limit AI within a WhatsApp conversation, use the WhatsApp guide instead. Its chat controls and trade-offs are different from the Instagram and Facebook choices here.
Is Meta AI safe? What is private and what you share
Leave a short record of what you changed
If you are helping a relative, let them make the choice and do the taps. Note the account, date and action in plain language: for example, 'submitted the public-data objection' or 'turned off Facebook photo suggestions'. Avoid the vague note 'AI off', which could mean several different things.
Check that the everyday part of the app still works as they expect. If they later see an assistant button, that alone does not tell you whether the data objection was recorded. Use the confirmation and the current official privacy information to check the preference that matters.
Helping older parents use AI safely: a practical guide for families
Sources
These sources support the guide. Tool settings, prices and policies change, so dated source boxes show when those details were checked.
- Meta: UK public information, AI training and the right to object (September 2024; checked 12 November 2026)
- Meta: Generative AI Privacy Guide (sign-in may be required)
- Meta: new AI features on Facebook, 15 June 2026
- Facebook Help: camera-roll suggestions (linked from Meta's June announcement)
AI helps with research and drafting. Dan checks the sources and is responsible for what is published.
How we check