OpenAI CEO Sam Altman has announced a major pivot in how the company’s AI video app Sora will handle copyrighted content: rather than allowing studios and creators to opt out of having their IP appear by default, OpenAI will now require opt-in permission and offer granular controls over how characters and works are used. In a blog update, Altman said rights holders will be able to specify usage limits or block inclusion altogether, and OpenAI intends to explore revenue sharing with those who opt in. This reversal comes after widespread backlash over unauthorized use of iconic characters and concerns that Sora’s earlier policy placed too much burden on creators. Meanwhile, OpenAI is also rolling out more user safeguards around how users’ own likenesses appear in generated videos, letting people restrict contexts, language, or placement of their digital “cameos.”
Sources: The Guardian, The Verge
Key Takeaways
– OpenAI’s shift to opt-in copyright control signals a retreat from a controversial default-permission model and aims to place control back with content owners.
– Granular settings will let rights holders dictate how, where or whether their IP is used in Sora videos, with a pathway toward revenue sharing for those who participate.
– Alongside IP changes, OpenAI is tightening user-facing safeguards — letting people limit where and how their AI likeness appears — as Sora faces scrutiny over misuse and deepfake risks.
In-Depth
OpenAI’s video generation tool, Sora, is stirring up the AI and entertainment worlds with a freshly announced policy reversal on handling copyrighted material. What started as a bold bet that Sora could rely on an opt-out system — where copyrighted characters would be included by default unless rightsholders demanded exclusion — has now pivoted to a more cautious, rights-sensitive approach. Sam Altman recently stated that OpenAI will deploy granular, opt-in controls that let rights holders control usage, including the option to block inclusion entirely. This move follows a wave of backlash from creators, studios, and legal observers alike.
Under the original plan, IP owners had to proactively submit opt-out requests to keep their characters or content out of Sora-generated videos. Many saw this as unfair: it put the burden squarely on creators to monitor and police use of their own work. That approach also risked flooding Sora with unauthorized renditions of beloved characters, raising copyright infringement fears and legal exposure. High-profile misuse cases — like Sora-generated videos featuring SpongeBob in edgy, unofficial scenarios — added fuel to the fire and intensified pressure. Some of these derivative clips even veered dangerously close to depicting illicit actions or using characters in contexts far removed from their original works.
Facing this backlash, Altman’s announcement indicates a tactical retreat. With the new system, rights holders must opt in for their work to be used, and OpenAI will introduce fine-grained rules over how characters are used — whether limiting contexts, forbidding certain behaviors, or blocking them entirely. OpenAI also plans to experiment with revenue-sharing models for creators who grant permission, to make participation more appealing and equitable.
In parallel, OpenAI is enhancing the controls users have over their own digital likenesses. Sora’s “cameo” feature lets users upload biometric data, enabling AI-generated video versions of themselves. Now, users can restrict where their likeness appears (e.g. disallow political content), block certain words or contexts, or impose custom filtering. These updates are intended to curb misuse and restore a measure of trust, especially given growing concerns about deepfake proliferation.
But the shift is not without challenges. OpenAI admits there will still be “edge cases” — content slips that bypass restrictions — and that the revenue model will require iteration. The legal landscape remains uncertain: even with opt-in controls, generating new content with recognizable characters may still spark disputes over derivative works or fair use boundaries. Studios and IP holders will need to engage proactively — reviewing permissions, drafting clauses, and monitoring usage. Meanwhile, OpenAI’s balancing act is delicate: it must maintain the creative promise of Sora, while ensuring it doesn’t trample creators’ rights.
In short, this about-face in policy reflects both technological ambition and caution. OpenAI is acknowledging that, for Sora to succeed sustainably, it must offer stronger protections, transparent consent, and fair incentives for rights holders. Whether this recalibration will placate critics or shift legal risk is still to be seen.