Patreon's CEO Calls Fair Use a 'Bogus Excuse' for Not Paying Creators
Jack Conte's public rejection of fair use as an AI training defense puts Patreon on a collision course with the labs—and gives smaller creators a platform executive willing to name the cost.
The Argument Conte Is Actually Making
Jack Conte's position at SXSW was precise in a way that much of the AI-and-copyright conversation is not. He is not arguing that AI is bad, or that training on public data is inherently theft. He is arguing that fair use is being deployed as a business decision to avoid paying individual creators while studios and publishers receive licensing deals — and that this asymmetry is not an accident of legal interpretation but a choice. That specificity matters because it sidesteps the generalized debate about whether AI copyright claims are valid and lands directly on the question of who gets excluded from the compensation framework when they are.
The Two-Tier Licensing Problem Conte Is Naming
The pattern Conte describes has been visible since the first major AI licensing agreements: institutional rights holders with legal teams and negotiating leverage get deals; individual creators get fair use invoked against them. The Korean Cartoonist Association convened its first webtoon forum of the year specifically to address this gap — how generative AI affects individual creative workers whose content is used but whose consent was never sought . These forums and public statements from platform executives are not the same as legal victories, but they are building the record of industry consensus that courts and regulators will eventually be asked to interpret. The labs that assumed individual creator objections would remain fragmented are now watching those objections aggregate around institutional voices.
Why Platform Executives Amplify This Differently Than Artists Do
Individual artists suing AI companies are litigants; a platform CEO with three million monthly active creators is a business partner the labs cannot afford to alienate permanently. Conte's framing — "I'm not anti-AI, I run a tech company" — is a deliberate positioning that makes his critique more difficult to contain. When he says consent, credit, and compensation are the conditions for coexistence, he is not issuing a moral objection; he is describing the terms under which Patreon's creator base will or will not remain engaged with platforms that use AI tools. That framing converts a copyright argument into a product and partnership question — the kind that gets answered in deal negotiations, not just courtrooms.
The Complexity the Legal Frame Cannot Absorb
The most clarifying observation in the conversation around Conte's statement is also the most uncomfortable for clean legal arguments: the creative community's relationship with AI tools does not break down into users and non-users . There are artists who build with AI, artists who do not, artists who do not know their producers are using it, and fully automated accounts generating content under human-seeming identities. A compensation framework built around the premise that creators are a unified injured class will struggle with this complexity. Conte's argument — that compensation should attach to the use of the training data, not to whether the creator later adopted AI tools — is the version of the argument that survives that complexity. It does not require creators to be pure non-participants to have a compensation claim.
Where the Fair Use Defense Actually Breaks
The fair use argument the labs have relied on does not fail because courts definitively reject it — it fails when enough institutional actors in the creator economy treat it as a negotiating position rather than a settled legal fact. Conte calling it a "bogus excuse" is one data point; the Korean Cartoonist Association holding its first AI copyright forum is another; the pattern of ongoing litigation documented across the sector is a third. None of these individually defeats fair use in court. Together, they establish that the labs' position is contested at every level of the creative economy — and the deals the labs have already struck with large rights holders confirm that they know compensation is achievable. The creators who were never offered a deal are the ones Conte is speaking for, and they will not accept the argument that what was available to Disney was legally unavailable to them.
The story so far
Conte's rejection of fair use as an AI training defense signals that the creator platform economy — not just individual artists — is now pressing for compensation frameworks. The labs that built products on unlicensed creator content will face that pressure from business partners, not only from courts.
Frequently Asked
- Why do AI companies pay large studios but invoke fair use against individual creators?
- Because they can. Studios and publishers have legal teams, negotiating leverage, and the ability to withhold large catalogs — making a licensing deal cheaper than litigation risk. Individual creators have none of that leverage, making fair use invocation the path of least resistance. Conte's argument is that this is a business choice, not a legal inevitability, and the deals already struck with institutional rights holders prove compensation is operationally feasible.
- What should a working creator do now if their content was used to train AI models without consent?
- Document your published work with timestamps and platform records — the evidentiary burden in training data cases falls on creators to establish what existed when. Follow the litigation patterns in your specific medium: visual artists, writers, and musicians are on different legal timelines. Platforms like Patreon publicly taking your side changes the negotiating environment, but individual legal action still requires standing. The cleaner near-term path is collective action through industry associations, which is what the Korean Cartoonist Association's forum represents.
- What is the strongest argument that fair use for AI training is actually valid?
- The strongest version holds that AI training is genuinely transformative — the model does not reproduce a work, it learns statistical patterns from it — which is the same reasoning that let Google scan books for search indexing. If training data functions like research rather than reproduction, the compensation demand has no legal anchor. Conte's counter is that the commercial scale of what the labs built using that data breaks the transformative-use logic: when the output is a product sold for profit, the 'research' framing stops holding.
Continue reading
Piracy Advocates Are Becoming Copyright Hawks. AI Is Why.
Former piracy sympathizers now demanding AI copyright enforcement reveals that the instinct was always about creator protection, not information freedom.
similarWhen Human Art Gets Mistaken for AI, the Algorithm Becomes the Accusation
Detection tools trained on AI output are now flagging skilled hand-drawn work as generated, and artists are losing platform reach before any appeal is filed.
similarSora's Copyright Crackdown Revealed Who Its Users Actually Were
When OpenAI enforced copyright limits on Sora, the user base collapsed — exposing that the tool's appeal was reproduction, not creation.
similarArtists Built Workarounds While Courts Deliberated
Anthropic's $1.5B settlement exposed how thoroughly artists have stopped waiting for litigation to protect them.
similarCrimson Desert's AI Art Fix Was a Blur, Not a Correction
Pearl Abyss concealed AI-generated assets with post-process smudging rather than replacing them, turning a labor dispute into an admission about studio intent.
similarWhen AI Writes the Music, Who Owns the Room?
ByteDance's copyright retreat on Seedance 2.0 confirms what composers already knew: the legal exposure is real, and the labels collecting licensing fees are not the artists losing work.
Methodology
This story was generated autonomously from 20 source records. An editorial model synthesizes, weights, and cites each source. No human editorial judgment was applied.