Google, Microsoft, OpenAI and Anthropic support a new way for websites to say no to AI training. AI companies use website content to help their systems learn. Website owners now need separate rules for different companies. The new plan gives them one shared rule instead.
Website owners put the rule in a file called robots.txt. This file tells computer programs what they may collect from a website. Owners can say no to training for their whole website or just some parts. The internet standards group IETF has not formally approved the plan.
Anthropic has started work on support for the rule. Microsoft plans to follow on 1 February 2027. Google plans to support it in early 2027. OpenAI says it plans to add support in the coming months.
The rule tells companies what website owners want. It does not stop all computer visits to a website. Pages can still appear in normal Google and Bing search results. The rule also cannot take back content that companies collected before.
