Microsoft and Anthropic join a shared opt-out that lets websites block AI training
One line in a website's robots.txt file would tell several AI companies at once not to train on its pages.
Microsoft and Anthropic say they will honor one shared signal that lets websites tell AI companies not to use their pages to train AI, joining Google and OpenAI. Microsoft laid out its plans in an Oct. 30 blog post.
Training means feeding huge amounts of text to an AI program so it learns to write and answer. Today, sites block each company's crawler, the program that copies web pages, one at a time in a file called robots.txt. The new signal adds one line that covers them all. It was proposed by Kevin Kelley of Anthropic at the Internet Engineering Task Force (IETF), a group that writes internet standards.
Anthropic already honors it, The Verge reported. Microsoft said it will start Feb. 1, 2027.
Microsoft will also let sites block grounding, when its AI pulls up a page to answer a question, as in its Copilot assistant. Blocking either one will not remove a site from Bing search results, Microsoft said.
Meta communications head Andy Stone voiced support but did not say whether Meta will honor it, The Verge reported.
Source: Microsoft Bing blog, Oct. 30, 2026, IETF draft by Kevin Kelley and The Verge, Oct. 30, 2026.