Microsoft and Google support unified AI training block mechanism

On October 30, Microsoft announced new rules regarding the use of website content in its AI products, giving resource owners distinct controls. On the same day, Google stated its intention to also support a general mechanism for banning AI training. These initiatives aim to establish a standardized approach for webmasters to manage how their content interacts with machine learning models.

• New rules separate restrictions into AI Training (model learning) and AI Grounding (using content to generate direct answers).
• Companies are aligning with an upcoming IETF standard allowing site owners to set universal rules across multiple AI providers.
• The changes will take effect in 2027, introducing a transition period for legacy tags while keeping pages visible in standard search results.

Separating training and answer generation

Microsoft proposes distinguishing between two ways AI utilizes website materials. The first method, AI Training, involves texts and images being fed into models so they can learn patterns and reproduce information independently. The second method, AI Grounding, occurs when the AI references specific site pages in real-time to answer user prompts while providing a source link.

Website owners can block AI Training by adding specific directives to their robots.txt file without affecting regular search indexing. For AI Grounding, a separate restriction can be applied using an HTML meta tag or an HTTP response header. If no preference is stated, Microsoft treats both methods as permitted by default.

Google's plans and IETF standardization

Google plans to integrate support for the general training ban throughout 2027, relying on the evolving IETF standard. If a site blocks training via the new directive, Google will consider training prohibited for its mechanism as well, though existing Google-Extended settings will retain priority. Google has not yet announced support for the AI Grounding restriction mechanism.

The IETF project is not yet an officially approved standard, and these new directives represent a declaration of site owner preferences rather than absolute technical protection against copying. Compliance entirely depends on the AI companies utilizing the content.

Implications for publishers and webmasters

Publishers now face a clear choice: allow AI to train on their materials, permit their use in generated answers, or restrict both approaches. However, Microsoft warns that blocking grounding will reduce the site's presence in AI-generated answers and decrease potential referral traffic from those services.

For digital businesses and content creators, this means adapting SEO strategies well ahead of the transition periods in early 2027. Replacing outdated tags like NoArchive and NoCache with the new standardized directives will ensure proper content governance while maintaining visibility in traditional search engine results.