Patreon is tightening restrictions on AI capture. The company stated that it had relied on robots.txt in the past to send a “no grab” hint to reptiles because some AI training reptiles did not comply with these rules. Now, Patreon starts to intercept the access directly.

From Tip to Direct Ban

According to Patreon, the company has expanded its cooperation with Cloudflare, using the latter ' s AI Crawl Control technology to update the platform ' s AI policy and implementation tools. Unlike before, which was mainly controlled by reptiles through robots.txt, this adjustment focused on the active banning of AI training robots.

According to the company, the upgrades were made because AI capture methods were more complex than in 2023. The early paid wall of Patreon made it difficult for a large number of creators to get their content by reptiles, but recently the platform has launched a new version of the first-page information stream and other discovery tools, such as Quips, with an increase in public visibility and new capture risks.

Test access attempt to zero

Patreon found in the tests that part of AI training reptiles to try to visit the platform on a weekly basis had dropped from thousands to zero. On this basis, the company judged that, in the past, a number of capture tools had continued to capture the content of the stations, despite robots.txt ' s restrictions.

According to Patreon, whether or not to consent to the use of content for AI training should not depend on whether the capturer is willing to follow the rules. This is also the main reason for the shift from “requests” to “direct interception”.

The index robot is still accessible.

But Patreon didn't completely block all the robots. The company indicated that robots used for page indexing and information collation would still be allowed to continue to access, and that they would be able to transport users back to Patreon.

This approach reflects the Platform ' s attempt to draw a line between content dissemination and training data control: on the one hand, the retention of search and detection capabilities and, on the other hand, the restriction of content being directed to training models.

Recently, a growing number of websites and content platforms have begun to address the training data capture by AI. Cloudflare has previously introduced tools to limit the use of AI robots and has access to a mechanism “Pay Per Crawl” that can charge the graber. Earlier this month, Cloudflare also adjusted its policy to intercept “mixed” reptiles, who are both indexed and trained, by default on advertising pages.

The director of the Patreon products, Drew Rowny, stated that, with the increased capacity and use of AI agents, creators should have a clearer decision on how the work should be used by AI companies. According to him, Patreon wanted the creator to retain control over the use of content while expanding its audience.