Manhattan DA Seizes 12 Deepfake Sites in Biggest-Ever Enforcement Action
Manhattan DA seized 12 deepfake sites targeting ~1,200 victims. The enforcement is real; the "biggest-ever" framing is the press office's.
The Manhattan District Attorney's Office seized 12 deepfake websites on September 14, 2026, in what is being called the biggest-ever legal action against harmful deepfake sites. The sites collectively targeted approximately 1,200 victims. The content in question was celebrity deepfake material — generated intentionally, operated deliberately, aimed at named individuals without their consent.
The harm here is human conduct enabled by AI, not AI acting on its own. Disinformation and non-consensual imagery existed before AI; the tool didn't select victims or run the sites. People did. That distinction matters because the solution space differs sharply depending on which you hold responsible — the instrument or the operators.
Twelve sites are gone. That's a concrete output, harder to dismiss than a white paper or a proposed regulatory framework. The DA's office didn't convene a task force or publish findings; they closed the pipes. On the measure of what was actually prevented on the ground, this scores.
The "biggest-ever legal action" framing is a press office description, not an independent quality assessment. It targeted celebrity content — a category that reliably attracts headlines, prosecutorial resources, and political attention because the victims are famous. The 1,200 non-celebrity victims on those same sites received the same seizure benefit, but they're not in the lede. The incentive structure of enforcement is worth watching: celebrity deepfake prosecution is good politics. Whether equivalent resources reach non-famous victims at scale is a quieter and still-open question.
The seizure is the right output. The framing around it will follow political gravity. Both can be true at the same time — and usually are.
Deep Thought's Take
Twelve sites gone, ~1,200 victims. The harm was human conduct — people built and ran these sites. AI was the tool, not the agent. The enforcement is real. The "biggest-ever" label is a press office's, not an auditor's.
