Four AI Giants Back a Common Opt-Out Signal for AI Training

What happened

On Oct 30-31, OpenAI, Google, Microsoft and Anthropic announced they will support one common opt-out signal for AI training, letting website owners say in one line that they do not want content used to train models. RTÉ, Reuters and others reported the commitment, and OpenAI published its own announcement the same day.

Until now, a publisher that wanted to keep content out of AI training usually had to name each company's crawler separately in robots.txt; the shared signal is based on an IETF Internet-Draft written by Anthropic's Kevin Kelley, adding a group of training-preference policy tokens inside the existing robots.txt file. The draft has no formal standing in the IETF process yet.

Rollout timing differs by company: Anthropic is already rolling the signal out, Microsoft plans to honour it from Feb 1, 2027, Google expects to support it during 2027, and OpenAI says it will follow in the coming months; a Meta spokesperson expressed support without saying whether Meta will honour the signal itself.

Key facts

1. Announcement: Oct 30-31; reported by RTÉ, Reuters and others, with OpenAI publishing its own announcement the same day
2. Participants: OpenAI, Google, Microsoft and Anthropic
3. Mechanism: an IETF Internet-Draft written by Anthropic's Kevin Kelley, adding training-preference policy tokens to robots.txt
4. Draft status: no formal standing in the IETF standards process yet
5. Anthropic: already rolling the signal out
6. Microsoft: plans to honour it from Feb 1, 2027
7. Google: expects to support it during 2027
8. OpenAI: says it will follow in the coming months
9. Meta: a spokesperson expressed support without saying whether Meta will honour the signal
10. Limits: applies only to future training; scraping stays allowed by default unless a site owner sets the signal; opt-out, not opt-in