Hollywood's Sora Betrayal: Agents Say OpenAI Knew Exactly What It Was Doing

Major talent agencies accuse OpenAI of calculated deception after Sora 2 launched without promised safeguards, sparking copyright chaos and deepfake scandals.

Three days after OpenAI released Sora 2, SpongeBob SquarePants was taking bong rips in AI-generated videos. Scooby-Doo was getting pulled over for speeding. The app had rocketed to a million downloads, and Hollywood's most powerful agencies were watching their worst fears materialize in real-time.

The breach of trust runs deeper than copyright violations. According to WME executives speaking to The Hollywood Reporter, OpenAI executives—including COO Brad Lightcap—had conducted "upbeat" pre-launch meetings where they assured agents that Sora 2 would protect intellectual property and client likenesses. Instead, the September 30 release unleashed what one WME partner called "a very calculated set of moves" by CEO Sam Altman. "They knew exactly what they were doing when they released this without protections and guardrails."

The timing matters. Weeks before Anthropic settled a $1.5 billion copyright infringement case in September 2025, OpenAI pushed ahead with Sora 2's launch. The app's "Characters" feature—which lets users insert themselves into any scene after a brief recording—became ground zero for deepfakes and unauthorized celebrity content.

"They're turning copyright on its head," says Rob Rosenberg, a partner at Moses Singer specializing in AI and copyright law. "They're setting up this false bargain where they can do this unless you opt out. And if you didn't, it's your fault."

The opt-out mechanism itself became a flashpoint. OpenAI initially required agents to individually notify them for each client who wanted protection—a logistical nightmare that one unnamed WME executive called impossible: "It's very likely that client would fire their agent... None of us would make that call."

By October 4, facing mounting criticism and a plummeting App Store rating of 2.8 stars, Altman published a blog post promising "granular control" over character generation. The language was vague. The timeline, vaguer.

Meanwhile, Altman had begun reframing the narrative. In public statements, he described the copyright-infringing content as "interactive fan fiction"—a creative expression rather than theft. The goal, he said, was to "make people smile" alongside making money.

WME wasn't smiling. On October 16, the agency released an internal memo opting all clients out of Sora 2 en masse. CAA called the situation "exploitation." The bridge between Silicon Valley and Hollywood, as one WME executive put it, had been "quite literally set on fire."

"How are you coming to the industry expecting partnership?" the executive asked, reflecting on the pre-launch meetings that now felt like theater.

The dispute reveals a fundamental clash over consent models. OpenAI's approach assumes permission until revoked. Hollywood's tradition demands explicit licensing before use. Neither side appears willing to budge.

What makes this different from previous AI copyright disputes is the allegation of deliberate deception. This isn't about technology outpacing regulation or unforeseen consequences. According to the agencies, this is about promises made in private rooms that evaporated the moment the app went live.

Agencies are exploring litigation options beyond copyright claims, potentially including bad faith negotiation. WME and CAA have implemented blanket opt-outs for all represented talent. App Store ratings suggest consumer backlash alongside industry anger. OpenAI's "granular controls" remain unimplemented three months after the promise. Competitive pressure is mounting as Google develops Veo 3 with potential YouTube integration.

The question now is whether OpenAI can rebuild trust faster than lawyers can file briefs. With Anthropic's recent settlement establishing a $1.5 billion precedent for copyright violations, the stakes have crystallized into hard numbers. OpenAI declined to provide specifics on when the promised safeguards would arrive or why agents were given different information in private meetings.

The most telling detail comes from that unnamed WME executive, contemplating the ruins of what seemed like a promising partnership: "You quite literally set the bridge on fire." In Hollywood, where relationships are currency, that's not a metaphor. It's an epitaph.

Source notes: OpenAI's "calculated moves" claim: src_pcmag_2025 SpongeBob/Scooby-Doo examples: src_futurism_2025 Million downloads in 3 days: src_roughcut_2025 "Upbeat" meetings with Lightcap: src_pcmag_2025 Anthropic $1.5B settlement: src_pcmag_2025 Characters feature details: src_openai_blog_2025 Rosenberg copyright analysis: src_futurism_2025 Agent opt-out impossibility: src_futurism_2025 Altman's "granular control" promise: src_pcmag_2025 "Interactive fan fiction" framing: src_pcmag_2025 WME memo and CAA response: src_pcmag_2025, src_futurism_2025 2.8 star rating: src_pcmag_2025