OpenAI has recently announced a significant policy change for its AI video generation model, Sora, by pausing the ability for users to create videos resembling the late civil rights icon, Martin Luther King Jr. This decision comes as a direct response to a request from Dr. King’s estate, which cited the generation of “disrespectful depictions” of his image by some Sora users.

The company, through a statement on its official newsroom account on X, clarified its stance, stating, “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.” This principle further extends to allowing authorized representatives or estate owners to formally request that a public figure’s likeness not be utilized in Sora cameos.

This restriction is notably imposed just weeks after the launch of Sora, OpenAI’s social video platform, which was designed to enable users to create realistic AI-generated videos featuring historical figures, friends, and individuals who consent to their likeness being recreated. The platform's introduction has ignited a fervent public debate surrounding the potential dangers associated with AI-generated videos and the crucial need for platforms to implement robust guardrails around such advanced technology.

The concerns voiced by Dr. King’s estate are not isolated incidents. Dr. Bernice King, Martin Luther King Jr.'s daughter, publicly appealed on Instagram for people to cease sending her AI videos resembling her father. Similarly, Robin Williams’ daughter has also made a plea for Sora users to stop generating AI videos of her late father. Reports from The Washington Post highlighted particularly egregious examples, including AI-generated videos of Dr. King making monkey noises and engaging in a wrestling match with fellow civil rights leader Malcolm X. Beyond these figures, crude videos resembling other prominent personalities like artist Bob Ross, singer Whitney Houston, and former President John F. Kennedy are also readily discoverable within the Sora app.

Beyond the representation of human likeness, Sora's launch has also prompted a cascade of questions regarding how social media platforms should manage AI videos involving copyrighted works. The Sora app reportedly contains numerous videos depicting popular cartoon characters from franchises such as SpongeBob, South Park, and Pokémon, raising intellectual property concerns.

In response to these burgeoning issues, OpenAI has been progressively introducing other restrictions to Sora since its debut. Earlier in October, the company revealed plans to provide copyright holders with more granular control over the types of AI videos that can be generated using their likeness. This move may be a direct acknowledgment of the initially negative reception from Hollywood regarding Sora’s capabilities.

Interestingly, this cautious approach with Sora stands in contrast to OpenAI’s seemingly more lenient stance on content moderation for ChatGPT. The company recently announced its intention to allow adult users to engage in “erotic” chats with ChatGPT in the coming months, highlighting a divergent strategy across its different AI products.

Ultimately, OpenAI appears to be actively grappling with the complex concerns inherent in AI video generation. Public statements from within the company, including CEO Sam Altman's admission of "trepidation" on Sora’s launch day, and Nick Turley, head of ChatGPT, asserting that the best way to educate the world about new technology is through its real-world deployment, underscore this learning process. It is evident that OpenAI is continuously refining its understanding of not only how to develop, but also how to responsibly distribute this powerful technology.