# Google, Microsoft, OpenAI and Anthropic back one AI training opt-out

> Four AI companies plan a shared robots.txt token for opting out of AI training.

*Site operators would block AI training with one robots.txt entry instead of one per provider.*

By Bez · WireRead
Canonical: https://wireread.com/news/google-microsoft-openai-and-anthropic-back-one-ai-training-opt-out

Google, Microsoft, OpenAI and Anthropic plan to support one shared robots.txt token that lets website operators opt out of AI training with a single entry. The companies said they have submitted a draft to the Internet Engineering Task Force (IETF), which maintains the Robots Exclusion Protocol.

According to [ppc.land](https://ppc.land/google-microsoft-openai-and-anthropic-back-one-ai-training-opt-out/), the token is named `__usage_ai-training`, and the draft is an individual Internet-Draft dated 30 October 2026. Site operators would need one robots.txt entry instead of one per provider. Secondary coverage reports that the IETF datatracker says the draft is not endorsed by the IETF.

According to [SiliconANGLE](https://siliconangle.com/2026/10/30/google-anthropic-and-others-to-support-new-ai-crawling-opt-out-standard/), the standard would not apply to models that perform exclusively non-generative tasks. The companies plan to add it to the Robots Exclusion Protocol.

Microsoft plans support in the coming months and is adding a separate AI Grounding setting for pages an AI model reads to answer a user's question. OpenAI plans support in the coming months, with similar controls. Google will follow sometime next year, SiliconANGLE said.

Secondary coverage gives more specific dates. Microsoft will honour the preferences from 1 February 2027, replacing the NoArchive and NoCache tags it has used since 2023. Google targets early 2027 and will read the token as a Google-Extended opt-out.

Meta has expressed support for the idea but has not committed to implementing it, SiliconANGLE said. SiliconANGLE gives no timeline for Anthropic, and Anthropic's own statement was not found. One roundup says Anthropic has already begun implementing support, while some coverage lists only OpenAI, Google and Microsoft as committing.

The token is a voluntary preference, not an access control, according to ppc.land. It does not replace GPTBot or Google-Extended rules. It applies only going forward, so content already crawled stays in existing training sets.

The IETF draft has to clear the AI Preferences working group next. The rollout dates are those listed above, and Anthropic's timing remains unconfirmed.

## Key takeaways

- Site operators can opt out of AI training with one robots.txt entry.
- The IETF draft must still pass through the AI Preferences working group.
- Microsoft will honour the token from 1 February 2027.

## Sources

- [Google, Microsoft, OpenAI and Anthropic back one AI training opt-out](https://siliconangle.com/2026/10/30/google-anthropic-and-others-to-support-new-ai-crawling-opt-out-standard/) — SiliconANGLE, 2026-10-30
- [Google, Microsoft, OpenAI and Anthropic back one AI training opt-out](https://ppc.land/google-microsoft-openai-and-anthropic-back-one-ai-training-opt-out/) — ppc.land, 2026-10-30