Ever wonder why AI companies fair use arguments sound hollow when they’re simultaneously cutting million-dollar licensing deals? Jack Conte, Patreon CEO and founder, called out this glaring contradiction at SXSW 2026. He terms it a “bogus” defense that conveniently excludes individual creators from the compensation that major publishers quietly receive. The intellectual property double standard is difficult to justify once the pattern becomes clear.
The AI Companies Fair Use Defense Falls Apart Under Scrutiny
AI companies have long justified training their models on copyrighted content without permission by invoking fair use doctrine. This legal principle allows limited use of copyrighted material based on four factors: purpose and character of use, nature of the original work, amount used, and market impact on the original.
But here’s where the AI companies fair use argument crumbles. They’re not applying it consistently.
The Selective Payment Pattern
So OpenAI struck licensing deals worth millions with The Associated Press and News Corp. Anthropic negotiated similar agreements with major publishers. If their fair use claims were legitimate, why pay anyone at all?
As Conte pointedly asked: “If it’s legal to just use it, why pay?” The answer reveals the fundamental weakness in AI companies fair use positioning. They recognize the legal vulnerability of their stance and pay large rights holders who can afford litigation, while denying payment to millions of individual creators who cannot.
Consider the stark reality: AI models have generated “hundreds of billions of dollars of value” for these companies, according to Conte, yet individual illustrators, musicians, and writers whose work contributed to this windfall remain uncompensated.
How AI Companies Fair Use Claims Mask Economic Discrimination
The selective application of fair use reveals something troubling—it’s being used as cost containment. Not legal principle. Disney, Condé Nast, Vox, and Warner Music all secured content licensing agreements. Individual creators? They get the fair use excuse.
The Resource Disparity Problem
This disparity isn’t accidental. Major publishers have legal teams and litigation budgets. They can challenge AI companies in court and win substantial settlements. Individual creators typically lack these resources, making them vulnerable to having their work used without compensation under questionable fair use claims.
In practice, I’ve seen this pattern repeatedly across different industries. Large corporations negotiate favorable terms while smaller players get offered “industry standard” excuses. The AI training data environment follows this same playbook perfectly.
What makes this particularly egregious is the scale. According to industry estimates, AI companies have ingested millions of creative works to train their models, yet creator payment AI systems remain virtually nonexistent for individual artists.
Why the AI Companies Fair Use Argument Contradicts Their Own Behavior
The contradiction becomes impossible to ignore when you examine specific licensing deals alongside fair use claims. OpenAI’s partnership with News Corp reportedly involves payments of $250 million over five years. That’s substantial money for content they simultaneously claim is legally free to use.
This behavior pattern suggests AI companies understand their fair use AI argument won’t withstand legal scrutiny from well-funded opponents. They’re essentially admitting through their actions that compensation is appropriate—just not for creators who can’t afford to sue them.
The Transformative Use Fallacy
AI companies often argue their use is “transformative” because they’re not reproducing content verbatim but processing it into new models. Yet this same logic could apply to any derivative work. Musicians sampling other songs, writers adapting existing stories, or artists creating collages all engage in transformative use—but still need permission or face copyright infringement claims.
The transformative argument becomes particularly weak when AI models can reproduce specific styles, techniques, or even recognizable elements from training data. That’s not transformation. That’s replication with extra steps.
Why Jack Conte’s AI Companies Fair Use Critique Lands Differently
Conte’s criticism carries weight precisely because he’s not anti-AI. “I run a frickin’ tech company,” he emphasized, positioning himself as someone who understands and embraces technological disruption. Patreon itself emerged as a solution to previous industry shifts that threatened creator income.
He compared AI to earlier disruptions creators navigated: the transition from music purchases to streaming, the rise of vertical video formats popularized by TikTok, and changes in social media distribution algorithms. Creators adapted to all these shifts. And they’ll adapt to AI too.
Adaptation Requires Economic Participation
But adaptation only works when creators can participate economically in the disruption. And streaming platforms pay artists, however little. Social media platforms offer monetization options. AI companies? They offer fair use arguments while paying select publishers millions.
“Societies that value and incentivize creativity are better for it,” Conte stated, connecting creator compensation to broader social good. When creators can’t earn sustainable income, cultural production suffers. And everyone loses.
Training Data Rights Represent the Next Battleground
The tension between AI companies fair use claims and selective licensing practices points toward an inevitable reckoning. As more creators become aware of how their work feeds AI development, legal challenges will multiply.
And several class-action lawsuits are already underway. Sarah Andersen, Kelly McKernan, and Karla Ortiz filed against Stability AI, Midjourney, and DeviantArt in Andersen v. Stability AI. Paul Tremblay and Mona Awad launched Silverman v. OpenAI over ChatGPT’s training data. The Authors Guild filed separately against OpenAI. But the outcomes will establish clearer precedents around training data rights—and AI companies’ selective licensing deals may prove their most damaging exhibit in court.
Publisher Licensing Deals Set Compensation Precedents
The millions AI companies pay publishers create a precedent that undermines their fair use defense. Courts will inevitably ask why some content creators deserve compensation while others don’t—especially when the legal theory supposedly applies equally to all copyrighted material.
This precedent strengthens individual creators’ legal positions. If Disney’s content deserves licensing fees, why doesn’t an independent illustrator’s work? The AI companies fair use argument can’t logically distinguish between them.
AI Companies Fair Use Debate: What Creators Should Do Now
Conte’s platform supports over 250,000 creators (as of Q1 2026), giving him unique insight into how AI impacts individual livelihoods. Patreon’s business model, enabling direct fan support for creative work, aligns perfectly with his compensation arguments.
A Patreon spokesperson confirmed their position reflects creator community feedback about AI content usage concerns: “At Patreon, our focus is on ensuring creators can build sustainable businesses, and that includes advocating for a future where creators are recognized and compensated for the value they bring, even as technology evolves.”
The path forward likely requires either extending licensing agreements to individual creators at scale or establishing legal frameworks mandating AI training data compensation. Conte’s emphasis on AI’s inevitability while maintaining creators must be “included economically” positions compensation as a creator economy issue, not an obstacle to technological progress.
Beyond Fair Use: Building Sustainable Systems
Smart AI companies will recognize that sustainable growth requires creator buy-in. Antagonizing the people who produce training data creates unnecessary legal and reputational risks. So collaborative approaches that compensate creators while advancing AI development benefit everyone.
But some companies are already exploring this path. Adobe’s AI tools offer opt-in compensation for contributors (Firefly, specifically). Getty Images negotiated licensing deals that benefit photographers. These models prove creator payment AI systems can work at scale.
When This Approach Has Limitations
Conte’s arguments face practical challenges that deserve honest acknowledgment. Compensating millions of individual creators for AI training data would be logistically complex and potentially expensive. Determining fair payment rates across different content types, usage levels, and creator tiers presents substantial administrative hurdles.
Some creators might prefer broader access to AI tools over direct compensation, especially if payments would be minimal when divided among millions of contributors. Overly restrictive licensing requirements could slow AI development or favor larger companies that can more easily afford complex rights clearance processes.
Alternative approaches like collective licensing organizations (similar to music performance rights societies) might offer more practical solutions than individual creator agreements. Or the challenge lies in balancing creator rights with innovation incentives while avoiding systems so complex they become unworkable for all parties involved.
The practical implication for individual creators is this: document your work, track where it appears in AI outputs, and connect with organizations like the Authors Guild or Artist Rights Alliance that are building collective legal capacity. Waiting for AI companies to voluntarily extend fair use licensing to individual creators hasn’t worked. The class-action route, the legislative route, and the collective licensing model are all moving simultaneously. The creators who engage now will shape the framework everyone else inherits.
Frequently Asked Questions
Do AI companies fair use claims have any legal merit?
Fair use doctrine does apply to some AI training scenarios, particularly for research purposes. However, commercial AI companies’ selective payment of some rights holders while claiming fair use for others undermines their legal position and suggests they recognize the weakness of blanket fair use defenses.
Why do publishers get licensing deals while individual creators don’t?
Publishers have legal resources to challenge AI companies in court and negotiate favorable terms. Individual creators typically lack the financial means to pursue litigation, making them vulnerable to having their work used without compensation under questionable fair use claims.
Could creator compensation requirements slow AI development?
Properly designed compensation systems could actually accelerate development by providing legal certainty and reducing litigation risks. Adobe and Getty Images have proven that creator payment models can work at scale without hindering innovation.
What’s the difference between AI training and other fair uses?
Traditional fair use typically involves limited excerpts for criticism, education, or commentary. AI training uses entire works to build commercial products that can compete with original creators—a much broader and more commercially significant use.
How might training data rights be resolved legally?
Current class-action lawsuits will likely establish clearer precedents. Courts will need to determine whether AI companies’ selective licensing deals undermine their fair use defenses and whether creators deserve compensation similar to what publishers receive.
