Patreon CEO Calls AI Firms' 'Fair Use' Argument "Bogus"

Patreon's CEO is calling out the "bogus" fair use argument used by AI firms, stating at SXSW that tech giants are profiting from creator content without compensation. His comments amplify a growing legal battle, with over 75 copyright lawsuits filed against AI model developers for using books, videos, and art as training data. This conflict highlights the widening gap between generative AI's explosive growth into a multi-billion dollar industry and the lagging standards for compensating creators.

Speaking at SXSW, Patreon CEO Jack Conte challenged the "fair use" defense from AI developers. He questioned why these firms strike expensive licensing deals with major media companies like Disney and Condé Nast while claiming the right to use independent creators' work for free (TechCrunch). While Patreon represents over 250,000 creators, its own policies have not yet formally addressed AI data scraping, revealing a disconnect between the CEO's stance and corporate action.

Patreon Chief: AI Companies Are 'Getting Away With' Using Creator Work for Free

Conte claims generative AI has built "hundreds of billions of dollars of value" on the backs of uncompensated creators. He proposes a two-part solution: enabling collective bargaining through platforms like Patreon and implementing metadata tags to facilitate micropayments for content use. While affirming he is pro-technology, Conte insists that creators must receive a share of the profits.

Patreon CEO Jack Conte argues that AI companies are exploiting a double standard. They claim "fair use" to scrape independent creator content for free, while simultaneously paying large sums to license content from major corporations, undermining the credibility of their legal argument for uncompensated data usage.

The legal system is now grappling with this issue. Since 2024, a wave of copyright infringement lawsuits has been filed by authors, artists, and video creators. A notable January 2026 class-action lawsuit accused Snap of scraping YouTube videos for its Imagine Lens model, mirroring claims against Meta and ByteDance (Copyright Alliance). Google, NVIDIA, and Anthropic are also defending against consolidated cases questioning their use of large-scale text and image datasets.

Early court decisions offer a mixed outlook. In one significant case, Anthropic agreed to a $1.5 billion settlement in Bartz v. Anthropic after a court determined that using pirated books for AI training does not qualify as fair use. This settlement, averaging around $3,000 per title, was substantial enough to trigger dataset purges.

Lawsuits surge while policy stalls

Currently, legal trackers identify around 50 active federal lawsuits in the U.S. challenging AI data ingestion. These cases typically allege direct infringement, removal of copyright information, and unfair competition. In response, AI companies argue their use is transformative, but progress in these cases has been hampered by disputes over access to source data and internal documents.

Meanwhile, the AI industry's focus remains on talent acquisition rather than data ethics. Hiring for AI and machine learning roles has surged 88% year-over-year, with salaries carrying a 12% premium. Reports indicate OpenAI offers engineers average equity packages of $1.5 million, a figure that stands in stark contrast to the zero compensation offered to the creators whose work fuels their models.

This creates a central conflict: billions in value are directed to AI engineers, while the original content creators receive nothing.

Conte sees a window of opportunity for creators, drawing parallels to the early days of music streaming before royalty structures were established. He suggests that a compulsory licensing scheme or a major settlement could be achieved if platforms like Patreon can unify creators into a powerful negotiating bloc. As it stands, the battle is characterized by rising legal costs and rapid AI development, leaving the crucial question of how - and if - creators will be paid for their data to be decided by courts, legislation, or collective action.

---

Why is Patreon's CEO calling the "fair use" defense "bogus"?

Jack Conte told the SXSW 2026 audience that AI firms pay Disney, Condé Nast and Warner Music millions in licensing deals while simultaneously claiming they can scrape independent creators for free. His sound-bite: "If it's legal to just use it, why pay?"
The contradiction, he argues, exposes the defense as a post-hoc excuse rather than a principled legal stance.

How much value have AI models taken from creators, according to Patreon?

Conte puts the figure at hundreds of billions of dollars, generated from the work of Patreon's 250 000+ creators without a cent in return.
He warns that today's scraping is tomorrow's business model and that failing to pay creators risks "hollow(ing) out the creative economy that feeds the tech sector."

What practical fix does Patreon propose?

Instead of individual lawsuits, Conte wants collective bargaining for the long-tail of creators plus metadata tags that could trigger micropayments whenever content is used for training.
The idea is to mirror the music-industry shift from piracy to streaming royalties, but in an AI context.

Are courts already siding with creators?

Early data is mixed.
- Anthropic paid $1.5 billion in Bartz v. Anthropic after a judge ruled that pirated books are not fair use; the sum works out to roughly $3 000 per infringed title.
- Yet in a parallel Meta case, the same court granted summary judgment on fair use for legally purchased books, showing no uniform precedent.
With 75-plus AI copyright suits filed since 2022 and about 50 still pending, Conte believes legislative clarity will lag, leaving platforms to negotiate or litigate case-by-case.

Has any AI company announced a royalty pool or opt-out plan for 2025-2026?

None surfaced in the research.
While OpenAI is on track to hand employees an average $1.5 million in equity per person to win the internal "talent war," no comparable fund exists for external creators whose data trains the models.
Conte's takeaway: until industry-wide compensation rails appear, platform pressure and courtroom losses are the only levers creators have.