According to CoinMeta, OpenAI, Google, Microsoft, and Anthropic have reached an agreement that websites only need to add two rules in `robots.txt`: `user-agent : __usage_ai-training` and `disallow : /`. These rules will inform these companies not to use their content for training AI. This rule is specifically aimed at AI training; it does not prohibit crawlers from accessing websites, nor will it necessarily affect AI from citing website content in its responses. Microsoft allows websites to individually prohibit AI training while still retaining the possibility for the content to be searched and cited by AI. Google plans to bind this rule with the existing Google-extended settings. Anthropic has already begun to be implemented, with Microsoft planning to make it effective in February 2027. Google is scheduled to be supported in 2027, and OpenAI will also be launched in the coming months.