The family of Robin Williams has reactivated his Instagram profile as a defense against AI deepfakes
The family of Robin Williams has reactivated his Instagram profile, inactive since his death in 2014, to use it to combat the spread of AI-generated deepfakes of his voice and likeness and replace them with authentic content.
The children of actor Robin Williams – Zak, Zelda and Cody Williams – have reactivated his Instagram profile, which had been inactive since his death in 2014. According to their statement, the profile is intended to be a “safe and trusted place” where shared stories, photographs and videos authentically reflect his legacy. Zelda Williams clarified on her own account that the aim is to combat the “widespread misuse of AI” involving his voice and likeness – the family will share verified authentic content as a counterweight to AI-generated imitations circulating online.
Previously, in 2025, Zelda Williams had asked fans to stop sending her AI-generated videos of her father and called it infuriating to reduce the life legacies of real people to a “vague resemblance in voice and appearance” as sufficient justification for creating content.
The source further states that the guardrails of some AI tools against generating the likenesses of public figures can be bypassed. According to the source, the now-discontinued Sora app from OpenAI was used to create videos featuring protected characters and celebrities, the Grok model still generates sexualized deepfakes of famous women according to Wired, and the Seedance model from ByteDance reached an agreement with Motion Picture Association over similar cases after users bypassed its weak guardrails to generate scenes featuring the likenesses of Hollywood stars. The source also mentions that the likenesses of celebrities such as Taylor Swift and Rihanna were misused to spread scams on TikTok.
Why it matters
The case demonstrates that, according to the source, technical restrictions (guardrails) against generating the likenesses and voices of real people in tools such as Sora, Grok or Seedance can be bypassed, leading to the spread of unauthorized content featuring both deceased and living celebrities – including misuse for scams. For surviving family members and public figures, this means a need to actively defend their legacy by sharing verified content; for companies operating generative models, it poses a concrete reputational and legal risk, as illustrated by the agreement between ByteDance and Motion Picture Association.
Two audiences, two different impacts
What this means
For individuals
The case is a reminder that AI-generated content featuring the likeness or voice of public figures (including those who have died) can circulate without the family's consent and should be approached critically, with authentic sources checked.
For a business
The case highlights the reputational and legal risks for companies operating generative AI models whose guardrails against generating celebrity likenesses can be bypassed – ByteDance reached an agreement with Motion Picture Association because of this, while the Grok model continues to generate sexualized deepfakes, according to the source.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.