OpenAI Halts MLK Jr. Video Generation Amid Racist Depictions
In a significant move to address ethical concerns, OpenAI announced Thursday it is now blocking users from generating videos featuring the likeness of the Rev. Dr. Martin Luther King Jr. on its Sora app. This decision comes after a disturbing trend emerged, with users creating offensive and racist depictions of the beloved civil rights leader.
The company confirmed that this action was taken following direct requests from the King Center and Dr. Bernice King, Dr. King’s daughter, who expressed strong objections to the misuse of her father’s image.
This incident marks yet another instance where OpenAI is compelled to strengthen the safeguards surrounding its powerful AI tools, which have frequently faced scrutiny for their potential for misuse and lack of sufficient ethical controls.
An OpenAI statement confirmed, “Some users generated disrespectful depictions of Dr. King’s image. OpenAI has paused generations depicting Dr. King as it strengthens guardrails for historical figures.”
The Sora app, developed by OpenAI—also known for ChatGPT—quickly rose to become the most downloaded iPhone app in the United States after its launch on September 30. It allows individuals to create sophisticated videos from simple text commands, leveraging advanced artificial intelligence.
However, this innovative technology was soon exploited. Users began producing clips featuring Dr. King and other prominent figures delivering fabricated remarks that were often vulgar, racist, or otherwise inappropriate.
Beyond historical figures, Sora has also been used to generate highly realistic videos of deceased celebrities. Examples include deepfakes of Amy Winehouse hosting a cooking show, Michael Jackson working at Walmart, and a fictional conversation between actors James Gandolfini and Robin Williams.
This led to public outcry, including a plea from Zelda Williams, Robin Williams’s daughter, who urged users to cease creating AI videos of her father. Dr. Bernice King echoed this sentiment, stating on social media, “I concur concerning my father. Please stop.”
In response to these concerns, OpenAI indicated that representatives of deceased public figures could now formally request to restrict the use of their likeness in Sora’s video generation capabilities.
“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,” OpenAI clarified on Thursday.
Experts express broader apprehension that platforms like Sora could become fertile ground for the propagation of disinformation and abuse. Other unsettlingly realistic videos created with the app have shown fabricated scenarios such as ballot fraud, immigration arrests, protests, crimes, and even attacks on city streets, blurring the lines between reality and simulation.
Furthermore, Sora’s ability to generate content featuring fictional characters has sparked significant debate regarding copyright infringement. Hollywood studios, in particular, have voiced concerns that Sora-generated videos may already be violating copyrights of various films, television shows, and characters.
OpenAI’s foray into Sora represents a strategic pivot towards building a social network, moving beyond its traditional focus on deep research in artificial intelligence. Cultivating a large user base for Sora could unlock substantial new revenue streams for the company, which currently operates without profitability.
(Reporting contributed by Adam Satariano)