What to Do If Someone Makes an AI Deepfake of You

Meta just shipped a new AI tool called Muse that can turn any public Instagram photo into a realistic deepfake — and Meta's own detection system doesn't reliably catch what it makes. If that made your stomach drop a little, here's the part that matters: doing this to someone without their consent is illegal in the U.S., and there's now a real, working process for getting it taken down. The federal TAKE IT DOWN Act requires platforms to remove a reported non-consensual deepfake within 48 hours, and the FTC has already started enforcing it against the biggest names in tech.

What just changed with Meta's new tool

Muse launched earlier this month. It lets anyone generate synthetic images using someone else's public photos as the starting material — no permission required, no check that the person in the source photo agreed to any of it. The tool that's supposed to catch manipulated images, Meta's own AI-detection system, doesn't reliably flag what Muse produces. That gap is the whole story: the tool got easier and cheaper to misuse in the same season enforcement against misuse finally got teeth.

Is there a law against AI deepfakes of women?

In the U.S., yes. The TAKE IT DOWN Act has been federal law since May 2025, and the FTC's enforcement window opened in May 2026. Under the law, any platform hosting user content has 48 hours to remove a reported non-consensual intimate image — deepfake or real — once a victim flags it. The FTC has already sent compliance warnings to more than a dozen platforms, including Meta, TikTok, X, Snapchat, Discord, and Reddit, citing penalties over $53,000 per violation. It isn't full protection. But it's the first time platforms have had an actual legal clock running against them instead of a vague community-guidelines promise.

The blueprint for this kind of law came from a young woman, not a legislature. In 2014, at 19, Olimpia Coral Melo filed the first version of what became Ley Olimpia in Mexico's Puebla congress, after an ex-partner shared an intimate video of her without her consent. It took seven years of organizing fellow survivors before Mexico passed it nationally in 2021, making non-consensual sharing of intimate images a crime with real prison time. In June, the Inter-Parliamentary Union pointed to her law as the model it's urging other countries to adopt, as parliaments try to catch up to the exact kind of AI-driven abuse Muse just made easier.

Why this matters for you

You don't have to be famous for this to touch you. UN Women has tracked sexualized deepfakes surging 550% in just a few years, and the overwhelming majority target women, not men. This isn't a celebrity problem — it's a "any woman with public photos" problem, which is most of us. It's the same undercurrent running through who's actually building the AI you use every day and what AI companionship apps get wrong about women: tools shipped fast, safety patched in after the fact, and women carrying the risk in the meantime. It shows up in the quieter costs AI passes on to us, too — this is just the sharpest edge of it.

One thing you can do today

If you ever have to deal with this, don't start with a lawyer — start with the platform's report form. Use the exact phrase "non-consensual intimate imagery" or cite the TAKE IT DOWN Act by name; naming the law moves your report past the general queue. Screenshot everything first — the image, the URL, the timestamp — before you hit report, because platforms sometimes pull content before you've saved proof it existed. If 48 hours pass with no action, you can file directly with the FTC at reportfraud.ftc.gov. And if it's someone you know who's targeted, believe her immediately and help her document it — that's often the hardest part to do alone.

Quote to sit with

"My silence had not protected me. Your silence will not protect you." — Audre Lorde

💌 Sarah