The Dead Can't Sue for Libel, But Deepfakes Raise New Ethical Alarms.

Zelda Williams, the daughter of the late actor Robin Williams, is sharing a deeply personal message with her father's fans.

"I'm asking you, please stop sending me AI-generated videos of my father. Don't think I want to see them or that I'll find them meaningful. I don't, and I won't," she expressed in an Instagram story posted on Monday. "If you have any respect, just stop doing this to him, to me, and to everyone else. It's pointless, a waste of effort, and I can assure you, it's NOT what he would have wanted."

It's likely no coincidence that Williams felt compelled to post this shortly after the launch of OpenAI's Sora 2 video model and the Sora social app, which enables users to create highly convincing deepfakes of themselves, their friends, and specific cartoon characters.

This capability extends to deceased individuals, who appear to be unprotected targets since, as noted by the Student Press Law Center, it is not illegal to defame the dead.

Sora restricts users from generating videos of living individuals—unless it's of themselves or a friend who has granted permission to use their likeness (what OpenAI refers to as a "cameo"). However, these restrictions do not cover the deceased, who can generally be recreated without hindrance. The app, currently available by invitation only, has seen an influx of videos featuring historical figures like Martin Luther King, Jr., Franklin Delano Roosevelt, and Richard Nixon, along with late celebrities such as Bob Ross, John Lennon, Alex Trebek, and, indeed, Robin Williams.

OpenAI's policy on generating videos of the dead remains ambiguous. For instance, based on TechCrunch's testing, Sora 2 will not produce videos of former President Jimmy Carter, who passed away in 2024, or Michael Jackson, who died in 2009, yet it did create content resembling Robin Williams, who died in 2014. While OpenAI's cameo feature allows living individuals to set guidelines for their appearance in videos generated by others—a response to earlier criticism of Sora—the deceased have no such control. I imagine Richard Nixon would be turning in his grave if he saw the deepfake I made of him endorsing police abolition.

OpenAI has not responded to TechCrunch's inquiry regarding the permissibility of creating deepfakes of deceased persons. Nonetheless, it's possible that generating deepfakes of late celebrities like Williams falls within the company's acceptable usage; legal precedent suggests OpenAI would likely not face liability for defaming the dead.

Join 10k+ tech and VC leaders for growth and connections at Disrupt 2025

Netflix, Box, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, Vinod Khosla — just some of the 250+ heavy hitters leading 200+ sessions designed to deliver the insights that fuel startup growth and sharpen your edge. Don't miss the 20th anniversary of TechCrunch, and a chance to learn from the top voices in tech. Grab your ticket before doors open to save up to $444.

Join 10k+ tech and VC leaders for growth and connections at Disrupt 2025

Netflix, Box, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, Vinod Khosla — just some of the 250+ heavy hitters leading 200+ sessions designed to deliver the insights that fuel startup growth and sharpen your edge. Don't miss a chance to learn from the top voices in tech. Grab your ticket before doors open to save up to $444.

"Seeing the legacies of real people reduced to 'this vaguely resembles them, so that's good enough,' just so others can produce awful TikTok content puppeteering them, is infuriating," Williams wrote.

Critics of OpenAI argue the company has been careless on such matters, which is why Sora was quickly saturated with AI clips of copyrighted characters like Peter Griffin and Pikachu after its launch. CEO Sam Altman initially stated that Hollywood studios and agencies would need to explicitly opt out if they didn't want their intellectual property used in Sora-generated videos. The Motion Picture Association has already urged OpenAI to address this, stating that "well-established copyright law protects creators' rights and is applicable here." Altman has since announced the company will reverse this stance.

Sora is arguably the most hazardous deepfake-capable AI model available to the public to date, given the realism of its outputs. Competing platforms like xAI are less advanced but have even fewer safeguards, allowing the creation of pornographic deepfakes of real people. As other companies narrow the gap with OpenAI, we risk setting a terrifying precedent if we treat real individuals—whether living or dead—as mere playthings.