Meta keeps launching AI features that assume consent: Why it keeps backfiring

Meta keeps launching AI features that assume consent: Why it keeps backfiring

Last week, Meta introduced a new Instagram feature that allowed anyone to create AI pictures through public accounts, without any prior permission being needed. The reaction was harsh enough for Meta to quickly pull the plug on this feature within days. This may sound familiar because this is the fourth time in less than two years that Meta introduced a product based on the same false premise about silent consent.

Digit.in Survey
✅ Thank you for completing the survey!

Also read: 10 monsoon gadget safety tips everyone should know

Take the case of the EU in 2024. In order to develop its AI models, Meta would use the public content of its Facebook and Instagram profiles. Everyone was automatically opted in and the only way to do something was to fill out an objection form afterwards. The privacy group NOYB called this practice a charade and claimed that according to the highest court in the EU, legitimate interest could not serve as grounds for Meta to violate people’s data protection rights, first for advertising, and now for AI development as well. Complaints have been filed in eleven different EU member states and the technical analysis revealed that the very form of opting out required logging in.

In contrast, Brazil adopted a more punitive approach. The National Data Authority issued an order instructing Meta to immediately cease using AI training on Brazilian data because there was a clear and present risk of irreparable damage to the users. The regulatory agency was not criticizing the idea of AI training. Instead, the issue that arose was the problem of consent design, since Brazilian users did not receive email notifications or in-app notifications like European users, and the eight-step opt-out mechanism buried within Meta’s Privacy Policy Center was in effect.

As 2025 arrived, the question of whether Meta informed users was no longer being considered, but rather whether there was anything legally wrong about having an opt-out policy at all. Meta continued to use the “legitimate interest” opt-out model, which was augmented by a problematic statement that opting out after the training of an AI model had already begun would not have any effect and would not result in the removal of information previously collected for training. This means that objection after AI training had already been done would be incompatible with a right to be forgotten.

Also read: Can AI make you a better cricketer? Str8bat co-founder Gagan Daga thinks so

And now the Instagram image feature. Same shape, compressed into days instead of years: public accounts opted in by default, no notification to the people whose likeness could be generated, a viral backlash, then a quiet reversal.

What strikes me writing about this pattern is how identical the fix is every time. Meta doesn’t move to opt-in by design. It waits for a regulator to threaten fines, or for a feature to go viral for the wrong reasons, and then rolls back just enough to make the immediate problem go away. The EU saga alone took roughly a year of pause, revised notice, and renewed objection just to land somewhere close to acceptable. Nothing about that process happened because Meta decided consent should come first. It happened because someone made the cost of not asking higher than the cost of asking.

That’s the true story, and it goes well beyond one feature on Instagram. Meta has the ability to make consent-first AI tools. It has done so in the past via gating where the government has compelled it to do so. Why does it not default to such a strategy? Simple. Opting in reduces the data set, and a reduced data set makes for a bad product. Each rollout is a gamble that the user base will go along and ignore whatever problem there is.

That gamble typically pays off. But sometimes, like last week, it doesn’t. The question isn’t if Meta will try this again. Based on the past couple of years, the answer to that is yes. The question is which product becomes too expensive, either financially or through backlash, to allow Meta to continue this strategy.

Also read: 3 months with Nvidia GeForce Now: I completed 5 AAA games and here is how it went


Vyom Ramani

Vyom Ramani

A journalist with a soft spot for tech, games, and things that go beep. While waiting for a delayed metro or rebooting his brain, you’ll find him solving Rubik’s Cubes, bingeing F1, or hunting for the next great snack. View Full Profile