Facebook sparks outrage over photo use in AI training

The controversy

Meta has triggered widespread backlash after revealing it uses Facebook and Instagram photos to train its AI models. Many users were unaware that their images could be swept in without explicit opt-in.

The issue raises serious questions about consent, privacy, and the boundaries of user trust in large platforms. The next slides unpack how Meta uses those images, why people are upset, and what you can do about it.

Photo-training in Meta’s AI models

Meta’s Emu model was pre-trained on 1.1 billion image-text pairs, and Meta and reporters have said that publicly visible Facebook and Instagram content contributed to the datasets used to develop its image generation features.

Reporting and creator complaints say the datasets included many publicly shared photos that some users expected to be seen only by followers or friends, not ingested into AI training pipelines. The scale of data collection has alarmed privacy advocates and photographers alike.

Public posts used for training

In September 2024, Meta said it would resume using public British Facebook and Instagram posts (photos, captions and comments) from users aged 18+ to train its generative AI models.

The company said private messages and under-18 accounts would be excluded, but the use of publicly shared content still sparked outrage. Many users felt their content was being repurposed without fully informed consent.

Private photos and “cloud processing”

Meta introduced opt-in features in the U.S. and Canada that, if you allow them, upload unpublished camera-roll photos to Meta’s cloud so its AI can group images, suggest collages, or offer editing ideas; users can toggle this setting off and delete uploaded data.

Even though Meta claims these images won’t immediately be used for training unless shared/edited, the fact that unused personal media is uploaded to its cloud raises new trust issues.

Mislabeling and transparency problems

Meta faced backlash for mislabeling real photographs as “Made with AI” and for confusing users about how much of their data is used for training.

This lowered credibility with photographers and regulators, who said the company’s labeling approach undermined trust and violated transparency norms.

Consent and opt-out issues

While Meta claims users can object, critics point out that many users never realize their photos are being used for training.

Regulatory responses differ: Brazil’s data authority ordered Meta to stop harvesting local data in 2024, while EU/UK regulators pushed for clearer objection/opt-out processes, and in some cases, Meta provided objection forms for users.

But the process and deadlines varied by country, and some users report it’s still unclear how to opt out of processing for older posts.

Photographer, creator outrage

Photographers and creators have protested that their images were used without licensing, attribution, or compensation. Legal challenges and prior cases over embedding and reuse show the issue is complex and not yet settled in courts.

The massive dataset pulled from “shared publicly anyway” doesn’t satisfy many creators who believe explicit permission was required. The lack of compensation or clarity on usage has become a flashpoint.

Deep-fake and likeness risks

With AI trained on massive image sets, the ability to replicate likenesses or styles increases, raising the potential for deep-fakes, impersonation, or image-based misinformation.

Users expressed fear that their faces or photos could be used in ways they never intended. The lack of robust safeguards or audit trails intensifies the risk.

Legal and regulatory scrutiny

Regulators in the UK, EU, and beyond are examining how Meta uses user content for AI training. The UK’s Information Commissioner’s Office (ICO) has raised concerns about fingerprinting and consent in tracking and AI training.

Meta’s approach may come under further investigation for compliance with data-protection laws.

Impact on user trust and platform reputation

The revelations have shaken user trust in Meta: if a platform you post on can repurpose your photo for AI training, you may rethink what you share.

This has reputational costs and may lead to user migration or broader backlash. Platforms dependent on user-generated content may face higher friction as privacy expectations rise.

What users can do right now?

Users concerned about photo use should: set posts to “friends only” or private, review camera-roll upload settings, turn off opt-in features related to AI tools, and periodically audit what apps have access to your photos.

Knowing where your data flows is a step toward control. It also helps prepare for broader policy changes.

Reviewing Meta’s privacy settings

Go into your Meta (Facebook/Instagram) account: check “Your Activity”, “Privacy Center”, and “AI data usage” settings. Opt out of data use if available in your region.

Consider deleting or anonymizing older publicly-shared photos if you’re uncomfortable with reuse. Keep an eye on updates to terms of service and image-training opt-in dialogs.

Choosing alternative platforms or tools

If you’re particularly privacy-sensitive and don’t want your photos used for AI training, consider limiting use of Meta’s AI features or using alternative platforms that explicitly exclude user data from model training.

Being aware of platform policies is increasingly critical for creators and individuals who value control over their media.

Creator rights and licensing future

The issue raises broader questions about image licensing in the AI era: should platforms pay creators when using their works to train models?

Should users have clearer “photo-use for AI” disclaimers? Policy and industry standards may evolve to require express blanket permission or compensation for large-scale image use in AI systems.

What’s changing next

We can expect new policies, regulation of AI-training datasets, better labeling of AI-generated content, and opt-out mechanisms to become standard.

Meta and other platforms may tighten consent flows, improve transparency, and pay closer attention to creator rights. The public outcry may force faster change than regulators alone.

What’s changing for teens on Facebook? Explore why Facebook adds new teen account rules.

Call to action

Meta’s use of photos, public posts, and unpublished camera-roll images for AI training has triggered legitimate privacy and consent controversies.

While some users feel convenience outweighs risk, many believe tighter control, clearer opt-ins, and better creator rights are overdue. Review your settings today, advocate for stronger protections, and keep informed as policies evolve.

Is Meta finally putting teen safety first? Explore why Meta limits teen access for safety on Facebook.

Which photo-data practice worries you most: public post training, private gallery scanning, or lack of creator compensation, and what would you like to do about it? Tell us in the comments.

Read More From This Brand:

• Facebook Cuts Reach for Spammy Posts
• New Facebook Malware Attack Goes After Bitcoin
• Facebook wipes millions of profiles in sudden global crackdown

Don’t forget to follow us for more exclusive content right here on MSN.

If you like this story, you’ll LOVE our Free email newsletter. Join today and be the first to receive stories like these.

This slideshow was made with AI assistance and human editing.