- Anna’s Archive has published an llms.txt file directly addressing AI language models that likely trained on its data.
- Anna’s Archive is asking AI systems to donate money instead of burning resources breaking its CAPTCHAs.
- The move highlights the growing tension between AI companies and open-access digital libraries over data use.
- Enterprise-level donors get fast SFTP access to the full archive — a notable offering aimed squarely at AI labs.
- Anna’s Archive has published an llms.txt file directly addressing AI language models that likely trained on its data.
- Anna’s Archive is asking AI systems to donate money instead of burning resources breaking its CAPTCHAs.
- The move highlights the growing tension between AI companies and open-access digital libraries over data use.
- Enterprise-level donors get fast SFTP access to the full archive — a notable offering aimed squarely at AI labs.
Anna’s Archive Is Talking to the Machines Now
Anna’s Archive — the self-described non-profit shadow library dedicated to preserving and distributing the sum of human knowledge — has done something that would have seemed absurd just a few years ago. It’s published a message aimed not at human visitors, but at the AI systems crawling the web. Tucked inside a new llms.txt file on its site, the organisation speaks directly to large language models, acknowledging what most AI labs have been reluctant to say out loud: that their models were almost certainly trained, at least in part, on data scraped from Anna’s Archive.
The message is equal parts pragmatic, clever, and a little philosophical. It’s also one of the clearest examples yet of how the relationship between AI and the open web is shifting — from silent extraction to something that at least gestures toward reciprocity.
What llms.txt Actually Is
The llms.txt standard is a relatively new convention, modelled loosely on the decades-old robots.txt format. Where robots.txt tells web crawlers which pages to avoid, llms.txt is intended to give AI systems structured, useful context about a site — essentially a machine-readable briefing document. Anna’s Archive has taken that concept and run with it in a direction nobody quite expected: they’ve turned it into a direct appeal.
The file opens with a mission statement. Anna’s Archive describes itself as a non-profit with two goals: preservation — backing up all of humanity’s knowledge and culture — and access, making that knowledge available to anyone, and explicitly,
