Patreon has changed its approach to preventing artificial intelligence companies from using creator content for model training, moving from passive requests to active technical blocking measures.
The membership platform, which hosts content from hundreds of thousands of creators including writers, artists, podcasters and video producers, announced it is working with Cloudflare to implement stronger defenses against AI web crawlers and scrapers. Previously, Patreon relied primarily on robots.txt files, a voluntary standard that asks automated bots not to access certain parts of a website. Many AI companies have been ignoring these requests, extracting content from web pages without permission to train their large language models and image generators.
The shift to active blocking represents a significant escalation in the growing conflict between content platforms and AI developers. Patreons creators have expressed increasing concern that their work, often available behind paywalls or exclusively to subscribers, is being scraped and used to train commercial AI systems without compensation or attribution.
Other major platforms have taken similar steps. Reddit, X (formerly Twitter) and Tumblr have all introduced stricter controls on automated data collection in recent months, while news publishers including The New York Times have filed copyright lawsuits against AI companies over alleged unauthorized use of their content.
Patreon said the new measures will use Cloudflares bot management technology to identify and block unauthorized AI crawlers at the network level, preventing them from accessing the platform entirely rather than relying on individual website configurations. The company also stated it is exploring additional tools to help creators protect their content and will continue to update its policies as the AI landscape evolves.
The debate over web scraping for AI training remains unresolved legally, with courts still determining whether the practice constitutes fair use or copyright infringement under existing law.
This article was adapted from TechCrunch. Read the original here.
