Microsoft has introduced separate controls for publishers who want to block their content from AI training and AI-generated search answers without losing their place in standard Bing results.
Previously, publishers had to accept all uses or drop out of the index entirely. The new system splits control into three independent parameters: AI model training, AI grounding in search responses, and traditional organic indexing. Each can be configured separately through distinct user-agent directives in robots.txt.
Generative answers in search reduce site traffic. Users get information directly in results and skip the click. For ad-supported media, that means lost revenue. Publishers can now choose whether to allow AI citations in exchange for brand visibility or block grounding and rely on organic traffic alone.
Google does not offer the same granularity. Blocking Google-Extended stops both model training and AI Overviews. There is no way to restrict grounding separately. Microsoft's approach gives publishers more flexibility and is an attempt to retain partners concerned about traffic loss from generative search.
For affiliate operators running traffic through content sites, the control matters. If a pre-lander appears in an AI answer, the user may never reach the offer. Blocking grounding preserves the chain: search, click, conversion. That is particularly relevant in verticals where every visit counts — nutra, dating, finance.
Blocking AI does not guarantee more clicks. If the search engine pulls an answer from other sources, users may not click through to any site. The decision should be based on CTR, time on site and conversion data. There is no signal yet that Microsoft favours sites participating in generative answers, but the algorithm could shift.
