The landscape of preference signaling for AI training

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The launch of powerful AI systems developed by large tech corporations has shifted the power balance between internet creators and those corporations. A series of efforts is underway to restore balance through contracts, licenses, and other technical signals for data crawlers. This article maps the most visible efforts to introduce preference signaling on the public web: Internet Engineering Task Force (IETF) AI Preferences, Creative Commons Preference Signals, Real Simple License, CloudFlare AI Crawl, and the TDM-AI Protocol.