One-sentence summary
On October 30, 2026, OpenAI, Google, Microsoft and Anthropic agreed to honor a common policy signal in robots.txt, letting sites block AI training in one step; the proposal is being discussed in the IETF AI-preferences working group.
Event Overview
- Time: Announced October 30, 2026.
- Companies: OpenAI, Google, Microsoft, Anthropic.
- Event: The four agreed to respect a common signal in robots.txt; once a site adds it, their AI crawlers must not fetch its content for model training.
Key Information
| Item | Detail |
|---|---|
| Mechanism | A common policy label in robots.txt |
| Status | Anthropic already adopted; Google and Microsoft plan 2027 support; OpenAI in coming months |
| Standards track | Discussed in the IETF AI-preferences working group |
| Google's existing control | Google-Extended (since 2023) remains |
| Meta's stance | Supports a common signal but has not committed to comply |
| Impact | Publishers can opt out once instead of per vendor |
Background and Impact
- Finer controls: Microsoft will offer separate opt-outs for training and anchoring; a site's choice does not affect its appearance in Bing results.
- EU alignment: Google says the mechanism matches the EU Code of Practice for general-purpose AI on machine-readable preferences.
- Significance: Gives publishers a unified entry point in the copyright and training-data fight against AI crawlers.
Official Sources
- Infobae: Sites can block OpenAI, Google and Anthropic AI training in one step
- OpenAI News: OpenAI News