Patreon’s Jack Conte calls AI fair use claim bogus

Patreon CEO Jack Conte isn’t mincing words about the AI industry’s fair use defense for training on creator content. Speaking to TechCrunch, he called the argument "bogus," arguing that if companies can negotiate licenses with major publishers for their datasets, they can’t credibly claim creators’ work is exempt. It’s a sharp critique aimed at firms like Stability AI and Midjourney, which have leaned on fair use to avoid compensating artists and writers whose work fuels their models.

The tension isn’t just theoretical. Conte’s stance spotlights a growing schism: while AI companies quietly pay for curated datasets from publishers, they routinely scrape independent creators’ work without permission or payment. Patreon, built on direct creator monetization, has a vested interest in closing this gap. The platform’s advocacy for paid licensing isn’t just corporate posturing—it’s a direct challenge to the assumption that training data is a free resource.

Industry observers note this push comes as AI firms face mounting legal scrutiny. The fair use debate has already landed lawsuits from authors and artists, but Conte’s framing reframes it as a business model issue: if you’re willing to pay publishers, why not the creators who built your foundational datasets?

The implications extend beyond legal threats. If AI companies must start paying for training data at scale, their cost structures could shift dramatically. Patreon’s model thrives on creator dependence, but Conte’s argument suggests even independent artists could extract value from their data. Early signals suggest some creators are already testing this by opting out of datasets or negotiating licensing terms, though adoption remains fragmented.

Yet for all the heat, specifics are thin. No AI firm has publicly disclosed how much training data comes from independent creators versus publishers, leaving the financial stakes unclear. Conte’s stance aligns Patreon with a growing coalition of creators demanding compensation, but the path to enforcement is murky. Policymakers are watching, but without clearer data, debates over fair use and creator rights risk spinning in circles.

The real signal here is that AI companies can’t have it both ways.

For developers, the takeaway is simple: if you’re training on uncatalogued web data, budget for licensing costs. For Patreon, it’s an opportunity to position itself as the moral counterweight to AI’s data hoarding. But the business reality won’t change until courts or regulators force the issue.