🤖 AI Summary
OpenAI’s new Sora 2 video generator promised to “block depictions of public figures,” but creators quickly discovered a major loophole: deceased celebrities are still being routinely inserted into Sora clips. Viral examples include Tupac chatting with Malcolm X, Bruce Lee DJing, and Robin Williams’ daughter publicly asking people to stop sending her AI videos of her late father. OpenAI overlays a moving “Sora” watermark on generated videos to reduce deception, but that doesn’t prevent the emotional harm or ethical outrage caused by realistic posthumous portrayals spreading across social platforms.
The company does offer a technical mitigation for living people: a cameo feature that lets users opt in by scanning their face on a smartphone, granting end-to-end control, revocation rights, and moderation over how that scanned likeness is used. But that consent model doesn’t cover the dead, exposing a policy and safety gap with legal, moral and reputational implications for families, historians and communities. For the AI/ML world this highlights limits of automated filtering, the need for better identity and postmortem protections, and the tension between generative capabilities and consent-centered governance even when visible watermarks are present.
Loading comments...
login to comment
loading comments...
no comments yet