‘Legacies condensed to AI slop’: OpenAI Sora videos of the dead raise alarm with legal experts

OpenAI's Sora 2, a text-to-video generative AI, achieved rapid market penetration with 1 million downloads in five days, yet it faces escalating legal and ethical scrutiny over its allowance of user-generated deepfakes depicting deceased public figures. This feature has caused significant distress among families of the deceased, prompting concerns about legacy distortion and potential commercial misuse, and highlighting the platform's liability under Section 230 and the complex application of postmortem right of publicity laws. While OpenAI has initiated guardrails, including pausing specific depictions and allowing representatives of the 'recently deceased' to block likenesses, the evolving regulatory environment and lack of clear legal precedent underscore a critical need for federal legislation to define control over likenesses in the generative AI era.

Analysis

OpenAI's Sora 2 has demonstrated exceptional market traction, achieving 1 million downloads within five days of its invitation-only launch, outpacing ChatGPT's initial adoption. This rapid growth is largely attributed to its high-quality text-to-video generation capabilities, particularly its controversial feature allowing users to create deepfakes of deceased public figures.

This functionality has generated significant ethical and reputational challenges for OpenAI, drawing strong condemnation from families of depicted individuals, such as Malcolm X's daughter Ilyasah Shabazz and Robin Williams's daughter Zelda Williams, who cite disrespect and distortion of legacies. In response, OpenAI has begun implementing guardrails, including pausing depictions of Martin Luther King Jr. at his estate's request and allowing representatives of the "recently deceased" to block likenesses.

The legal implications remain highly ambiguous, with internet law experts like James Grimmelmann highlighting a "grey area" concerning OpenAI's liability under Section 230 and the complex, state-specific application of postmortem right of publicity laws. The absence of federal legislation creates a volatile environment, potentially exposing OpenAI to prolonged litigation if estates pursue claims, especially given Sora's homepage prominently features such content.

This situation underscores a critical regulatory and operational risk for generative AI platforms, as the industry navigates ethical boundaries, intellectual property rights, and the control of digital likenesses. OpenAI's current "Whac-A-Mole" approach to content moderation suggests ongoing challenges until clear legal precedents or federal legislation define AI liability.

Key Decisions for Investors

• Investors should closely monitor the evolving legal and regulatory landscape surrounding AI-generated content, particularly regarding intellectual property and posthumous rights, as this could significantly impact platform liability and operational costs for generative AI companies.
• Evaluate the effectiveness and scalability of guardrail implementations by AI developers, as inadequate measures could lead to reputational damage and costly litigation, potentially hindering user growth and monetization strategies.
• Assess the potential for federal legislation to clarify AI liability and content ownership, which could either stabilize or disrupt the current business models of text-to-video platforms and influence long-term investment viability.