Starting December 16, Meta will begin using publicly available content from its platforms to train and improve its artificial intelligence systems. This applies to Meta owned services including Instagram and Facebook.

This policy does not mean your posts are republished or attributed to you. It means they may be ingested as training data to improve model performance.

While some online commentary exaggerates the implications, there are legitimate data governance and user control concerns, particularly for public accounts.

What Counts as Public Content

According to Meta’s published explanations, public content may include:

• Public posts and captions

• Publicly shared images and videos

• Writing style and linguistic patterns

• Voice data derived from videos or audio content

• Metadata e.g. timestamps, engagement signals, and contextual information

Public accounts and public posts are explicitly within scope. Private accounts and private messages are treated differently under Meta’s stated policies.

What “Training” Actually Means

AI training does not involve copying or reposting your content verbatim. In practice, it means your data may be used to:

• Learn general language patterns and syntax

• Improve image recognition and generative models

• Refine tone, style, and contextual understanding

• Abstract facial features, voice characteristics, and writing style into model parameters

• Separate content from its original audience or intent, causing loss of context

If content has already been used for training, deleting it later does not retroactively remove it from trained models. At that point, user control is effectively lost.

This process is not inherently malicious, but it shifts control away from users and toward the platform.

Private Messages and AI Interactions

Meta has publicly stated that it does not use private messages with friends and family to train its generative AI models.

However:

• Meta does acknowledge that safety and integrity scanning occurs in messaging systems for spam, fraud, and abuse detection

• This scanning is performed by automated systems, not humans.

Even when done for security purposes, automated scanning is imperfect and prone to false positives.

Interactions With Meta AI Features

Meta has stated that interactions with its AI features may be used to:

• Personalize content and recommendations

• Inform advertising signals

• In the EU, contribute to training and improving AI models

This includes chats with Meta AI inside Instagram or Facebook.

Do not use Meta AI as a therapist, confidant, or journal. Avoid discussing sensitive, emotional, or personally identifying topics in AI chats within Meta apps.

EU Specific Disclosures

Meta has stated that in the EU it trains AI using:

• Public posts and comments shared by adults

• Interactions with Meta AI, including questions and queries

Meta also claims that sensitive topics are excluded from ad targeting. That does not eliminate the broader data processing concerns associated with AI systems operating at scale.

Children and Personal Images

Any content you post publicly, including photos of your children or family members, may be processed as training data.

Meta states that it only uses material from public accounts.

A. Meta is not a company you can trust.

B. Once data is ingested into large scale training pipelines, granular control is functionally lost. Training datasets are replicated, transformed, abstracted, and integrated across multiple systems and vendors. At that point, enforcing strict access control is almost impossible.

Even if a company claims policy level restrictions, there is no realistic mechanism for a user to independently verify:

• Where the data is stored

• How long it is retained

• Which internal systems can access it

• Whether third party contractors or vendors are involved

The loss of user control begins at ingestion, not publication. That is the core risk.

Your strongest protections are:

• Making your account private

• Opting out of AI training where legally available

Privacy settings reduce exposure but are not perfect.

How to Opt Out in the EU and UK

In the EU and UK, users have the right to object to their personal data being used for AI training. Meta states it will honor these objections.

Option A: In App Instagram Path

1. Open Instagram

2. Go to your profile

3. Tap the menu icon

4. Open Settings and activity

5. Select Privacy Centre

6. Locate a section such as Generative AI at Meta

7. Tap Your right to object

8. Submit the objection form

9. Use the email associated with your account

Suggested message:

“I object to my personal data being used for AI training.”

If the In App Path Errors

Use the direct forms instead.

Instagram objection form:

help.instagram.com/contact/767264225370182

Facebook objection form:

www.facebook.com/help/contact/71287672071558

Opening these links in a browser where you are already logged into the correct account significantly reduces technical errors.

Meta has also stated that EU users will receive an in app and email notification with a direct link to the objection form. If you receive that email, it is often the fastest and least frustrating route.

FOR THOSE IN THE UNITED STATES AND THE REST OF THE WORLD.

Go to Settings:

Scroll all the way down to About

Inside About select Privacy Policy

Click 3 lines on top right corner.

Select other policies and articles

Expand and click “How Meta uses information for generative Al models and features”

From there, it gets a little tricky but stick with it. Expend Privacy and Generative AI option.

Scroll all the way down to find this link

Click the link and scroll down to the little box you need to check.

Enter the name your Instagram account is under, initials work fine too. Fill out the form, and in the two boxes where it asks for information, simply write: “Please do not use any of my data to train your AI tools.”

At this stage, you may run into issues. The system might return a vague “something went wrong” message or claim your request does not meet the criteria. It is very frustrating. Meta has made opting out almost impossible, and only a very small number of people have succeeded. I contacted everyone at Meta whose email I could find about this, and several journalists and others did the same. We were told the issue would be fixed. I have not tried again since I already completed it on my account, but if you have not done it yet, please do.

Yes, it is frustrating. Yes, it is confusing. And no, that is not an accident. The friction is intentional. The harder it is to find and disable, the fewer people will bother, which conveniently allows Meta to continue extracting value from your data at scale.

If you are an artist, designer, model, writer, or simply someone who values privacy, you do not want your work feeding systems you do not control. Especially not Meta’s.

Meta does not operate on caution or restraint. It operates on growth, speed, and monetization. That combination has never produced careful data stewardship in the history of technology.

Once your content is ingested, you lose visibility, leverage, and recourse. You will not know how it is used, what it is combined with, or where it surfaces later.

If you get “something went wrong,” message do this checklist

1. Update Instagram to the latest version

2. Try the form in a mobile browser (Safari or Chrome) instead of the in app browser

3. Try desktop browser

4. Log out, log back in

5. Turn off VPN if you are using one

6. If you have multiple accounts in Accounts Center, double check you are logged into the correct one before submitting

There are numerous other applications you use daily that already collect data, which is subsequently utilized to develop or enhance their AI systems. Platforms such as LinkedIn, Gmail etc. Prior to their policy updates, I published step-by-step opt-out guidelines for my Instagram subscribers. So most of you should already be good with that, but I will incorporate or update these analyses here as well when time allows.

For now, the priority is to complete the Meta AI objection where it is available to you

Again, if your profile is public and primarily focused on your personal life, especially if it includes photos or videos of your little children, please make the account private. That single change materially reduces exposure. I promise you, you’ll be glad you did one day.