An individual Internet-Draft defines a shared training-preference token, and three companies plan support on different timelines; no simultaneously effective standard exists yet.
OpenAI, Google and Microsoft back a shared AI-training preference signal
2026-10-30Verified announcementWATCH
An individual Internet-Draft defines a shared __usage_ai-training token inside robots.txt. OpenAI says it plans to honor the signal in coming months, Google in 2027 and Microsoft from February 1, 2027. The draft is not an IETF standard or endorsement, support is not yet simultaneous, and compliance remains voluntary.
What this means for you
For web publishers, a common machine-readable signal could reduce provider-specific configuration. It does not block crawling, delete prior training data or create enforcement, so existing crawler controls should remain until implementations are live and tested.
Evidence & related material
Technical details & scores
| Document Status | individual Internet-Draft; no formal IETF standing or endorsement |
|---|---|
| Token | __usage_ai-training in ordinary robots.txt groups and path matching |
| OpenAI | planned support in coming months; existing controls remain |
| planned support in 2027; maps opt-out to Google-Extended unless explicitly set | |
| Microsoft | planned support from 2027-02-01 with transition from NoArchive/NoCache |
| Limits | voluntary preference signal, not access control, retroactive deletion or legal enforcement |
| Time Basis | Official pages give 2026-10-30 without time of day; discovered after the prior checkpoint during the scheduled gap review. |
Technical impact 2
Engineering value 3
Adoption signal 2
Maturity 2
Verification cost 2
Risk 3