The "Do Not Train on Me" Button: OpenAI, Google, Microsoft and Anthropic Back One robots.txt Signal
Key data at a glance:
Signal name: __usage_ai_training (written in robots.txt with Disallow: /)
Announced: October 30, 2026 by OpenAI, Google, Microsoft; Anthropic authored the draft
Draft: "Machine-Readable Training Preference Signal", IETF AI Preferences working group, by Anthropic's Kevin Kelley; published October 30, 2026, expires May 3, 2027
Timelines: Anthropic already honoring; OpenAI "coming months"; Microsoft February 1, 2027; Google aiming 2027
Scope: training data only; does not restrict crawling, search indexing, or citations in AI answers
Status: voluntary; draft not yet a ratified IETF standard
Sources: himiao.ai, David and Goliath AI, Tencent Research Institute, threerouter
Top comments (0)