Context
Google Search Central announced plans to support a shared robots.txt policy token for AI-training preferences, drawing on the IETF AIPREF work. Google aims to implement support in 2027, while preserving its existing Google-Extended-specific control.
What changed
Google says a shared policy-token opt-out will be interpreted as disallowing Google-Extended when Google-Extended is not explicitly specified. Explicit provider-specific Google-Extended rules will continue to operate alongside the shared signal. This is announced intended behavior, not a completed rollout.
Why it matters
The proposal would allow publishers to express broad training-use preferences without automatically treating them as equivalent to search-crawling or answer-grounding controls. Google says websites that specifically allow Google-Extended are eligible for grounded appearances and clicks in Gemini Apps.
Visibility OS interpretation
Visibility OS interpretation: AI content-use governance is moving toward shared purpose-specific signals with provider overrides. Audit training permissions, crawler access, grounding eligibility, actual source selection, and observed citations as separate dimensions. No ranking or citation benefit is established.
What remains uncertain
When in 2027 will Google implement and enforce this control? Will other providers converge on the same semantics? What effect, if any, will different Google-Extended settings have on observed Gemini Apps grounding or referrals? The IETF proposal is not a final standard.
Evidence claims
On October 30, 2026, Google announced plans to support a shared AI-training policy token in robots.txt in 2027; absent an explicit Google-Extended instruction, Google plans to treat a shared-token training opt-out as a Google-Extended disallow, while retaining provider-specific overrides.
Related historical records
OpenAI commits to shared robots.txt AI training preference signal
That record related to this event.