OpenAI blocks AI videos of Martin Luther King Jr. : NPR

OpenAI has blocked users from making videos of Martin Luther King Jr. on its Sora app after the civil rights leader's heirs complained that “disrespectful images” were being circulated.

Since the company launched Sora Three weeks ago, hyper-realistic fake videos of King saying crude, offensive or racist things quickly spread across social media, including fake videos of King stealing from a grocery store, running from police and perpetuating racial stereotypes.

Late Thursday, OpenAI and the King Estate released joint statement saying AI videos depicting King are being blocked because the company is “strengthening guardrails for historical figures.”

OpenAI said it believes there is a “strong interest in free speech” in allowing users to create artificial spoofs of historical figures, but that the estates should have full control over how these likenesses are used.

The Sora app, which is still available by invitation only, has taken a shoot-first-aim-later approach to security fences, raising alarms among intellectual property lawyers, public figures and disinformation researchers.

When someone joins the app, they are encouraged to video themselves from different angles and record their speech. Users can control whether others can make fake videos of them, which Sora calls “cameos.”

But the app allowed people to take videos of many celebrities and historical figures without explicit consent, allowing users to create fake footage of Princess Diana, John F. Kennedy, Kurt Cobain, Malcolm X and many others.

Cristelia Garcia, a professor of intellectual property law at Georgetown Law, said OpenAI's actions only after a complaint was filed against King's estate were consistent with the company's approach of “asking for forgiveness, not permission.”

“The AI ​​industry seems to be growing very quickly, and 'first to market' seems to be the currency of the day (due to the contemplative, ethical approach, of course),” Garcia told NPR via email.

She noted that rights of publicity and defamation laws vary by state and may not always apply to deepfakes, meaning that “there may be some minor legal downsides to just letting things run their course until someone complains.”

Although the ability to control one's likeness depends on where one's property is located, some states have strong protections, e.g. Californiawhere a public figure's heirs or their estate own the rights to the image for 70 years after the celebrity's death.

Days after releasing the Sora app, OpenAI CEO Sam Altman announced changes to an app that gives copyright holders the option to choose to have their images depicted using AI, rather than having that image allowed by default.

However, the families of some deceased celebrities and public figures have criticized OpenAI for allowing the depiction of vulgar, unflattering or compromising behavior.

After videos of Robin Williams took over social media, Zelda Williams, the late actor's daughter, asked the public to stop making videos of her father. “Please just stop sending me videos of my dad's AI,” she wrote in an Instagram post, adding that “this is NOT what he wanted.”

Bernice King, daughter of a civil rights activist, agrees: letter to X: “Please stop.”

Hollywood studios and talent agencies have also expressed concern that OpenAI submitted the Sora app without the consent of copyright holders.

This is a similar approach to how the company developed ChatGPT, which has absorbed a huge number of copyrighted content without approval or payment, before eventually entering into licensing agreements with certain publishers. This approach caused a wave copyright claims.