Google Is Spending Billions to Make AI Look Good — And Disney Has the Power to Say No

Google is prepared to spend roughly $40 million for the rights to generate a single copyrighted character using its AI models — and potentially well into the billions for a full studio library — but as of August 31, 2026, not one major Hollywood studio has signed on the dotted line. The Los Angeles Times broke the story, citing three sources familiar with the conversations, and the silence from Disney, Universal, and Warner Bros. Discovery tells you more about the deal dynamics than any official statement would.

What's Converging

The timing of Google's push isn't accidental. Across the last eighteen months, AI companies have rushed to convert their legal adversaries — the studios, record labels, and publishers who sued them over training data — into paid partners. The implicit theory is that a licensing deal turns a litigation threat into a revenue relationship and, more importantly, confers a kind of cultural approval that scraped data never could. OpenAI pursued the same logic with Disney and walked away empty-handed before an agreement was signed. Runway signed a deal with Lionsgate in 2024; as of today, nothing has shipped from that partnership.

That track record matters because it establishes a pattern: the headline deal is easy to announce in theory and hard to finalize in practice. Multiple reports indicate that the sticking points consistently involve union obligations, talent consent, and the question of who bears liability when an AI model produces something that looks too much like a specific actor. SAG-AFTRA's contract with the studios requires notification whenever an AI licensing agreement is executed — meaning the first real signal of a closed deal is more likely to surface through a union bulletin than a press release.

Google's own moves have been unusually diverse. Earlier in 2026, DeepMind — Google's AI research division — made a $75 million investment in A24, the indie studio behind some of the decade's most culturally resonant films. That investment didn't buy training data; it bought proximity to a brand that younger audiences actually trust. The pattern suggests Google understands that what it's really acquiring isn't pixels and dialogue — it's association.

The Specific Development

The LA Times report published August 31 describes Google executives approaching representatives at Disney, Universal, and Warner Bros. Discovery with pitches centered on Gemini, Google's flagship AI model family. The ask: let Google train on your library, and receive both a substantial upfront payment and a share of advertising revenue generated when AI-produced content featuring your IP runs on YouTube. The rough math, according to one person familiar with the discussions, puts a single character license at approximately $40 million. Scale that to a hundred characters and the total crosses into multiple billions.

What's striking about the YouTube component is that it came packaged with a separate, thorny ask. According to the LA Times, some companies were approached about participating in a YouTube likeness-detection tool — a system designed to flag AI-generated content that reproduces celebrities' faces or infringes on copyrighted material. The catch, per the reporting, is that participation required waiving the right to sue Google in connection with that program. The studios apparently declined that condition, and no major studio has filed the SAG-AFTRA notification that a finalized AI deal would require.

The Jingletree analysis of this situation frames it plainly: Google benefits from these deals with little downside, while each studio absorbs distinct reputational and legal risk. The Spider-Man: Brand New Day AI artbook controversy is the illustration closest at hand — a film that grossed roughly $2 billion faced genuine public backlash when an AI-generated artbook was associated with the production. That reaction, on a project with every commercial advantage, shows how quickly audience goodwill can curdle when the AI association is perceived as careless or exploitative.

What's Likely Next

The most important variable to watch over the next 30 to 90 days isn't whether a deal gets signed — it's whether any studio formally notifies SAG-AFTRA. That notification is the observable tripwire. AI Weekly's coverage notes that a Google editor familiar with the situation expects the first real signal to come through union channels rather than corporate announcements, precisely because studios are sensitive to how talent and crews would react to a publicly declared AI licensing partnership.

Our read is that Google's negotiating position is weaker than the money figures suggest. The studios understand something the AI companies have been slow to internalize: the legitimacy problem isn't technical. Gemini can already generate compelling images and coherent video. What it cannot generate is the public trust that comes from a Disney or Pixar stamp of approval — and that trust is exactly what has eroded fastest over the past two years of AI disruption hype. An LA Times source put it directly: for Google's tools to be usable at the scale of actual film and television production, its models need to be fine-tuned on content and metadata that only Hollywood studios possess. That's a dependency, not a partnership of equals.

The precedent that matters most here isn't the Lionsgate-Runway deal or the collapsed OpenAI-Disney talks — it's how this plays politically inside each studio's talent relationships. The first major studio to sign a Google deal will become the test case for whether AI licensing can coexist with union trust. Given that SAG-AFTRA members just finished negotiating AI protections into their contracts, the studio that moves first absorbs the reputational cost of being the first to monetize those same protections away. The money Google is offering is real. So is the risk. The studios are, for now, declining to decide which one is larger.

Sources