Reported information
Four AI providers back a shared training opt-out signal
Google, Microsoft, Anthropic and OpenAI announced support on 30 October 2026 for a common robots.txt signal to opt content out of AI training, The Verge reports.
The IETF draft uses the token `__usage_ai-training` with Allow and Disallow rules. It concerns using content to train generative models; rules controlling crawler access remain separate.
OpenAI plans support in the coming months and says existing controls remain in force. The voluntary mechanism remains a working document, rather than a final standard or technical barrier ensuring every collector respects preferences.