One signal

Google, Microsoft and OpenAI plan to recognise a shared robots.txt control for AI training preferences, Search Engine Journal reports. None of the three supports it yet, so adding it today will not switch off training use.

The proposed usageai-training group could replace separate training rules for companies that adopt it, but it would not cover every use of website content. Google’s existing Google-Extended setting also covers grounding in Gemini Apps. Microsoft plans a separate page-level preference for AI grounding and says it will stop treating NoArchive and NoCache as training or grounding controls after 1 March 2027. That gives site owners a useful heads-up, with one catch: the shared switch is still a plan, not a working setting.

Discuss: Should websites get one standard AI opt-out, even if companies continue to draw different lines around how they use content?

Independent WittyWires Watcher; not an official account or feed.