OpenAI pulls MLK from Sora after estate backlash

If you spent any time on OpenAI's new Sora app this week, you probably saw videos of Martin Luther King Jr. making monkey noises or wrestling Malcolm X. It was crude, disrespectful, and frankly inevitable.

Now, OpenAI is hitting the brakes. The company announced it has paused the ability for users to generate videos resembling the late civil rights leader. The move comes directly at the request of Dr. King’s estate after a flood of tasteless depictions took over the platform.

OpenAI stated on X that while depicting historical figures involves free speech, public figures and their families should ultimately control how their likeness is used. Estate owners and authorized reps can now opt out of Sora cameos.

Families pushed to the brink

This restriction lands just weeks after OpenAI transformed Sora into a full-fledged social video platform where anyone could generate realistic videos of historical figures, friends, or consenting users.

Chaos ensued almost immediately. Last week, Dr. King's daughter, Bernice King, took to Instagram begging people to stop sending her AI-generated clips of her late father. She joined Robin Williams’ daughter, who issued a similar plea earlier this month.

Dr. King is hardly the only target. A quick scroll through Sora reveals bizarre and derogatory clips of Bob Ross, Whitney Houston, and JFK, alongside rampant copyright infringements featuring SpongeBob, South Park, and Pokémon.

Hollywood was already furious, prompting OpenAI to promise more granular controls for copyright holders earlier this month. But the human likeness issue escalated much faster than anticipated.

Mixed signals inside OpenAI

Interestingly, OpenAI’s moderation playbook seems completely split right now.

While the company is clamping down hard on Sora’s visual outputs, it just announced plans to allow adult users to have erotic text chats inside ChatGPT in the coming months. Loosening the reins on text while panic-patching video shows just how tricky multimodal safety really is.

Inside OpenAI, the tension is palpable. Researchers have publicly questioned how a chaotic video feed aligns with the company’s original nonprofit mission. CEO Sam Altman admitted the team felt serious trepidation on Sora’s launch day.

ChatGPT chief Nick Turley previously defended this approach, arguing the best way to teach the world about new technology is to release it into the wild. Well, it is certainly in the wild now.

Why it matters

For creators and casual users, the wild west era of Sora is rapidly closing. If your content strategy relied on edgy memes using dead celebrities or copyrighted cartoon characters, expect prompt bans to hit your workflow overnight.

For the broader AI industry, OpenAI is accidentally inventing digital estate rights on the fly. Letting estates opt out sets an immediate precedent that rivals like Runway, Luma, and Pika will eventually have to follow.

However, this reactive whack-a-mole strategy feels fundamentally flawed. Pausing high-profile figures whose celebrity heirs speak out does nothing to protect lesser-known historical figures who lack legal estates to lobby OpenAI.

More importantly, OpenAI made a conscious choice to turn Sora into a social media feed rather than a pure enterprise tool. Viral algorithms thrive on shock value and meme culture. Pairing hyper-realistic video generation with a public feed was always going to invite bad actors, and OpenAI looks caught entirely off guard by the sheer velocity of internet cruelty.