Contents
- The Short Answer
- What Meta's AI Actually Trains On
- Turning Off AI Ad Targeting, Step by Step
- Phone-Level Permissions Worth Checking
- Filing a Formal Objection to AI Training
- Regional Differences: Where This Actually Works
- What Opting Out Doesn't Do
- Stopping AI-Generated Images of You
- Things No Source Confirms Yet
- Managing Meta AI Chat History
- FAQ
- What's Still Unknown
Meta will review an objection to using your Instagram data for AI training. Reviewing isn’t approving — and even an approved objection only stops future use; nothing already folded into a training run comes back out. What follows is the fine print other write-ups on this topic tend to skip: which menu holds which switch, and which country’s opt-out actually has teeth.
The Short Answer
Meta has used public posts from adult users to train its AI since 2007. That part isn’t new.
What changed in June 2026 is the paperwork: a setting called “Activity from other businesses” now governs how your data feeds AI-driven ad targeting, while a separate Privacy Center form lets you formally object to the training itself. They’re 2 different switches, and flipping one doesn’t touch the other.
The toggle people actually care about sits somewhere else: under “Sharing and reuse,” it controls whether someone can generate an AI image of your face using your handle — without notifying you. It’s not easy to find.
Since December 16, 2025, Meta has also used your conversations with Meta AI as a signal for ad personalization. That’s a separate stream from the training question, and it’s easy to miss.
What Meta’s AI Actually Trains On
Before touching any setting, it’s worth knowing what’s actually being fed into the model. Meta’s own disclosures put 3 categories in scope:
- Posts you’ve shared publicly
- Photos, including their captions
- Comments you’ve left
Direct messages are excluded, unless you’ve shared them yourself with Meta AI inside a chat. So the exposure here is mostly about what you’ve posted publicly, not what you’ve said in private.
One thing you can’t control: photos other people post of you. If a friend tags you in a picture, that image sits under their privacy settings, not yours.
Data already folded into an existing training run can’t be pulled back out afterward. Opting out stops future use — it doesn’t undo past use.
Turning Off AI Ad Targeting, Step by Step
This setting is the easiest of the bunch to find. On iPhone:
- Open the Instagram app
- Tap your profile, then the 3-line menu in the top right
- Go to Accounts Center
- Select “Activity from other businesses”
Simple enough. Except it isn’t always in that spot.
On other accounts, the same control shows up under a different label instead: Ad preferences → Ad settings → Activity information from ad partners. Same function, different path. None of the 5 sources checked for this piece flag that split — so if you go looking for a menu item that doesn’t exist on your account, you might assume the feature was removed. It wasn’t. It’s just filed elsewhere.
If you’re in the EU, there’s an extra lever: Ad preferences → Ad settings → Ad experience, where you can select “less-personalized ads” — a regional option not confirmed to exist for US or Asia-Pacific accounts.
One more thing that comes up in comment sections: Meta states, officially, that it does not use your phone’s microphone to listen to conversations for ad targeting. Treat that as an official claim, not a verified one — no source here tested it.
Phone-Level Permissions Worth Checking
Instagram’s in-app settings are one layer. Your phone’s own permission system is another, and it’s worth a separate pass.
On iPhone:
- Settings → Privacy & Security → Tracking
- Find Instagram in the list
- Toggle tracking off
On Android:
- Settings → Apps → Instagram
- Tap Permissions
- Adjust individually (location, contacts, microphone, and so on)
Neither of these stops Meta from using content you’ve already posted publicly. They limit tracking and cross-app data collection. Different problem, different fix — don’t confuse the two.
Filing a Formal Objection to AI Training
Ad targeting is the easy fix. Training is the one buried deepest, and this is the step that actually addresses it.
The current path:
- Go to your profile, tap the menu (≡)
- Select “More info and support”
- Open Privacy Center → Privacy topics
- Find the “Generative AI” section and tap “Get started”
Or skip the app entirely and go straight to privacycenter.instagram.com.
From there: choose “Information that you’ve shared on Meta products,” enter your email address, and submit. Total time, according to one source that timed it: about 5 minutes.
There’s an alternate route described elsewhere — Settings & Privacy → Privacy Center → “Right to Object” form, where you enter an email and a reason for the objection before submitting. Whether this is a separate form or the same one under a different label isn’t fully clear across sources. If one path doesn’t work, try the other.
For context: in 2024, UK users were pointed to this through a notification inside the Facebook or Instagram app itself — an “object” link that opened a short form. Email required, explanation optional.
Meta’s own stated position, quoted directly: if an objection is approved, the company will not use your public Facebook or Instagram information “to develop or improve generative AI models” going forward.
That phrase — “going forward” — is the entire limit of this opt-out, in 3 words.
Regional Differences: Where This Actually Works
This table decides whether anything above actually applies to you: it lays out where each country stands.
| Region | Legal basis | What happens |
|---|---|---|
| EU / UK | GDPR-based formal opt-out rights | High likelihood of approval |
| United States | No federal privacy law requiring it | Objection can be filed, approval not guaranteed |
| India | No enforceable opt-out right in place | Limited, effectively unavailable |
| South Korea | Not addressed in any source reviewed | Unconfirmed |
The EU and UK situation is the strongest, because it’s backed by law, not by Meta’s goodwill. The US situation is the opposite: you can submit the form, but nothing legally requires Meta to honor it, and none of the 5 sources reviewed here give an approval rate or a processing timeframe.
South Korea isn’t mentioned in any of the 5 sources used for this piece. Not “opt-out unavailable” — just absent. If you’re in Korea looking for a definitive answer, none of the current coverage gives you one.
What Opting Out Doesn’t Do
Here’s what an approved objection actually changes — and what it doesn’t.
- The process is not automatic. Meta reviews and approves objections; it isn’t a switch that instantly takes effect.
- Meta’s operating assumption, as reported, is that public data is fair game for AI training by default — objection is the exception you have to actively request, not a baseline right you start with.
- Even where it works, it doesn’t work for everyone or block everything. One source describes it plainly: opting out “cannot fully block” AI use of your data.
- The process itself is still described as cumbersome — 2 years after it first launched in 2024.
- There’s a structural criticism worth repeating: under GDPR, sensitive personal data is generally supposed to require opt-in consent, not opt-out objection. Critics have flagged that Meta’s approach here runs the opposite direction — you’re in by default, and objecting is on you.
None of the sources reviewed give a number for how many objections get approved, or how long approval takes. If that data exists, Meta hasn’t published it anywhere these 5 sources found.
Stopping AI-Generated Images of You
Of everything in this guide, this toggle carries the sharpest real-world consequence — and it has nothing to do with ads.
The path: Instagram profile → menu → “Sharing and reuse” → find the section labeled “Allow people to reuse your content on Instagram and with AI features at Meta” → turn off the toggles for both Posts and Reels.
Here’s why it matters. Leave that toggle on, and anyone can tag your Instagram handle to generate an AI image using your likeness — and you won’t be notified when it happens. That’s not a hypothetical risk. It’s the direct mechanism behind identity misuse, phishing setups, and fabricated images circulating without your knowledge.
This isn’t an abstract concern, either. In July 2026, Meta’s “Muse Image” feature reportedly let Meta AI generate images referencing public Instagram accounts, and Meta pulled the feature soon after amid criticism — though the exact shutdown date and Meta’s internal explanation for it aren’t confirmed by the sources checked for this piece.
The fix for that specific feature was Meta removing it — not a setting you controlled. The “Sharing and reuse” toggle is a separate, ongoing control; check it regardless of whether Muse Image ever affected you.
Things No Source Confirms Yet
These search terms show up again and again around this topic, and it’s worth being direct: none of the 5 sources checked for this piece address them.
- Whether Instagram labels AI-generated content or AI-run accounts with a visible tag — unconfirmed
- Whether there’s a dedicated “AI creator” account category — unconfirmed
- Whether Instagram has banned any category of AI use outright — unconfirmed
- Background details on how Instagram’s AI systems are built or trained beyond what’s listed above — unconfirmed
- An exact deadline for filing an AI-training objection — not specified in any source, including the original TechCrunch coverage of the UK process
If you’ve seen a claim about any of these presented as fact elsewhere, treat it skeptically until it points to a primary source.
Managing Meta AI Chat History
2 commands handle this directly, typed straight into a Meta AI conversation:
/reset-ai— deletes a specific AI conversation/reset-all-ais— deletes all AI conversations at once
Neither of these is documented as retroactively removing anything already used for training. They clear the chat log going forward, which is a privacy control, not a training opt-out.
FAQ
Does turning off “Activity from other businesses” stop Instagram from training AI on my posts? No. That setting controls ad targeting signals. Stopping AI training use requires the separate Privacy Center objection process described above.
If Meta approves my objection, does it delete data already used? No. Sources describe the block as forward-looking only. Data already folded into an existing training run isn’t pulled back out.
Is the opt-out guaranteed to work if I file it? Only where it’s backed by law. In the EU and UK, GDPR gives you a formal right, and approval is likely. In the US, you can file the same form, but nothing requires Meta to approve it, and no source here gives an approval rate.
What’s Still Unknown
Whether Korea has any opt-out right at all, and what legal basis it would rest on, isn’t addressed anywhere in the 5 sources used here. There’s no published deadline for filing an AI-training objection, and nobody has published an approval rate or a processing-time figure for these forms.
Whether “Activity from other businesses” has rolled out identically across every region — as opposed to the alternate “Ad preferences” layout other accounts still show — isn’t confirmed either. If Meta has published harder numbers on any of this since these sources went live, that hasn’t surfaced yet.
Sources:
- Engadget, “How to limit Instagram from using your data for AI and ads,” published 2026-08-22
- TechCrunch, “Hey UK, here’s how to opt out of Meta using your Facebook and Instagram data to train its AI,” published 2024-10-03
- Malwarebytes, “Turn off this Meta setting before someone generates AI images of you,” published 2026-07-09
- DeleteMe, “How to Opt Out of Instagram AI Training,” published 2026-06-16
- Izoate, “How to Opt Out of Meta AI Training (2026): Stop Meta Using Your Data,” published 2026-04-02
Image credits
- Caption This Week 7 (9/12 Thru 9/16) by NASA Goddard Photo and Video, licensed under BY (via Openverse)
- ‘Honeycombs’ and Hexacopters Help Tell Story of Mars by NASA Goddard Space Flight Center, licensed under BY (via Openverse)
- Caption This Week 8 (9/19 Thru 9/23) by NASA Goddard Photo and Video, licensed under BY (via Openverse)
- Caption This Week 2 (8/8 thru 8/12) by NASA Goddard Photo and Video, licensed under BY (via Openverse)
- Coelioxys afra, f, right, J Smit, Netherlands_2021-11-23-15.30.28 ZS PMax UDR copy by Sam Droege, licensed under PDM (via Openverse)
- Caption This Week 4 (8/22 thru 8/26) by NASA Goddard Photo and Video, licensed under BY (via Openverse)
- Caption This Week 6 (9/5 thru 9/9) by NASA Goddard Photo and Video, licensed under BY (via Openverse)