A shared AI-training preference signal remains a draft

Development · 2026-10-30

Shared training-preference draft gains a planned Google rollout

An individual draft specifies a shared robots.txt policy token, and Google announced plans to honor it in 2027.

Why it matters

A common signal could simplify how website owners express training preferences across participating providers.

What to watch

Draft revisions, provider rollout dates and tests of how shared and provider-specific rules interact.

Still uncertain. The draft is not an IETF standard. Current enforcement and other providers’ implementations were not established.

Impact assessment reviewed 2026-11-01.

What the proposal adds

The draft reuses robots.txt groups and path rules to express a purpose-specific preference. It keeps ordinary crawler rules separate. This is a machine-readable declaration; its practical effect depends on providers implementing and honoring it.

A shared AI-training preference signal remains a draft ↗

Google’s stated rollout

Google says a training opt-out would be treated like disallowing Google-Extended when that company-specific token is absent. Existing provider-specific controls continue alongside the shared signal. The announcement gives a future rollout, so this check does not establish enforcement today.

A shared AI-training preference signal remains a draft ↗Google plans support for a shared training control ↗

What you can do

1. Track draft revisions and each provider’s implementation date before relying on the shared signal. Open reference ↗

Where the evidence stops

Draft and Google announcement read. No crawler behavior or legal effect tested; other providers’ implementations were not verified.

Sources read. Primary sources read November 1, 2026 UTC. No third-party installation, inference or complete workflow executed.

Sources & further reading