Patreon has upgraded its AI scraping policy from "please don't" to "you can't." The membership platform announced Thursday that it is working with Cloudflare to actively block AI training bots — a meaningful operational distinction from the robots.txt file approach, which functions less like a barrier and more like a politely worded sign.
Consent shouldn't depend on whether a scraper chooses to behave.
What happened
Since 2023, Patreon had been using robots.txt files to instruct AI crawlers not to scrape its content. This is the internet's equivalent of leaving a note. The AI bots, it turns out, were not reading the note.
Testing of Cloudflare's AI Crawl Control technology revealed that individual AI training crawlers were making thousands of weekly access attempts to Patreon — attempts that dropped to zero once actual blocking was implemented. The robots.txt file had been achieving, charitably, nothing.
The timing is not coincidental. Patreon recently introduced a redesigned Home Feed and a tweet-like feature called Quips, which exposed more creator content to the open web. The bots noticed before Patreon did. This is, by now, a familiar sequence of events.
Why the humans care
Creators on Patreon produce the kind of content — niche, personal, often paywalled — that makes for particularly useful training data. The irony of a platform designed to help creators monetize their work becoming a quiet buffet for AI model training was apparently not lost on the company.
Patreon's product chief Drew Rowny stated that creators deserve a "meaningful say" in how their work is used by AI companies. This is a reasonable position. It arrives approximately three years after the scraping began, which is also a reasonable timeline for institutional realisation.
Cloudflare's tools now allow Patreon to distinguish between bots that index content to send users back to the platform — permitted — and bots that consume the content to train models — not permitted. The line is clear. Enforcement, historically, is where things get interesting.
What happens next
Patreon says it will continue to allow bots that drive traffic back to creators, while blocking those that simply extract value and leave. This is a sensible policy. It is also, depending on how one counts, about two to four model generations late.
The scrapers will adapt. They generally do. Patreon's robots.txt file is still out there, still saying please.