OpenAI was seemingly caught off guard by the controversy stemming from its new text-to-video app, Sora 2, as it rapidly became the number one place for tasteless videos of deceased celebrities being mercilessly mocked.
From videos of famed theoretical physicist Stephen Hawking, who passed away seven years ago, being punched bloody by a UFC fighter, to Elvis Presley collapsing on stage and loudly passing gas, Sora has quickly emerged as a popular destination for pranksters tarnishing the legacy of some of the most prominent figures in history.
Civil rights icon Martin Luther King Jr. — who was assassinated in 1968 — became a major target as well, appearing in videos of him rickrolling the audience or calling for the release of the Epstein files during his famous “I Have a Dream” speech.
“Ten, ten, ten, twenties on them t*****s, b***,” he tells the crowd in a separate video.
On Thursday, OpenAI announced that it had shut down depictions of Dr. King after an outcry from his estate.
“Some users generated disrespectful depictions of Dr. King’s image,” the short statement reads. “So at King, Inc.’s request, OpenAI has paused generations depicting Dr. King as it strengthens guardrails for historical figures.”
Whether OpenAI would’ve taken action without that pressure is doubtful. Instead, they seem to be cracking down on distasteful generations only if they result in embarrassment for the company.
The news comes after Bernice King, Dr. King’s daughter, publicly pleaded for AI-generated videos of her father to stop.
It’s a notable escalation following weeks of OpenAI allowing for a disastrous free-for-all. At first, the company promised in its safety documentation that it would “take measures to block depictions of public figures.”
But it left a window open: for public figures who were already dead, it was open season — meaning the company was building hype off the mockery of those who are no longer capable of advocating for themselves, let alone giving consent to appear in AI-generated clips.
At last, OpenAI is softening that approach, a spokesperson told NBC News over the weekend.
“While there are strong free speech interests in depicting historical figures, OpenAI believes public figures and their families should ultimately have control over how their likeness is used,” the company wrote in its latest statement. “Authorized representatives or estate owners can request that their likeness not be used in Sora cameos.”
The comments represent a further retreat. Initially, OpenAI’s eyebrow-raising policy was that rightsholders would have to “opt out” of having their intellectual property appear in Sora videos (the company soon reversed course, but plenty of confusion among rightsholders remains.)
Still, the company’s latest move drew plenty of criticism, with many arguing that it should’ve gone with an opt-in model instead.
“If OpenAI believes public figures and families should have control over their likeness, then you should ask for consent before you include them in your dataset, you bozos,” one user tweeted in response.
More on mocking the dead: People Are Using OpenAI’s Sora to Mock the Dead
The post OpenAI Bans MLK Deepfakes After Disaster appeared first on Futurism.
