Updated November 5, 2026. Google's recent changes concern three different things: how AI answers link to websites, how site owners measure visibility, and how they control inclusion or model training. They should be evaluated separately.
Links Are Becoming More Visible in AI Answers
On October 15, Google said it had increased inline links in AI Mode and AI Overviews and was launching an update to place them higher in responses. It also began expanding timely-article carousels to more locations and languages, and made related-article links more prominent with thumbnails. A dedicated section for official websites is still being tested; clickable creator and publisher names on AI Mode video results are also a test. These are display changes, not a promise of more traffic to any particular publisher. Read Google's October product announcement.
For publishers, the practical check is whether the pages linked in representative answers actually answer the underlying question. Compare what appears in AI answers with your own landing pages and referrals before drawing conclusions about impact.
Search Console Now Shows AI-Feature Impressions
Google's dedicated generative AI performance reports reached websites worldwide by August 31. They cover impressions in AI Overviews, AI Mode, and generative AI features in Discover, with breakdowns such as pages, countries, and dates; device data is available for Search. The reports show where URLs appeared, not a query-by-query log of citations or a direct measure of resulting revenue. See Google's Search Console announcement.
Record a baseline for the pages that matter to your business, then compare AI impressions with referrals and on-site outcomes. An impression alone does not show that a reader clicked through.
AI Search Inclusion and AI Training Have Different Controls
The Search generative AI control in Search Console lets site owners include or exclude their links and content from AI Overviews, AI Mode, and generative AI features in Discover. Google says this control reached websites worldwide by August 31. Excluding a site removes its eligibility for links, impressions, and traffic from those features; it does not control model training or ordinary Search results.
For training, Google already offers Google-Extended. On October 30, it announced plans to support a shared AI-training preference in robots.txt in 2027. Google says an explicit Google-Extended instruction would take priority over the shared preference. That Google support is a plan, not a live rollout. Anthropic says it began recognizing the shared training signal on October 30 and that its specific ClaudeBot rules take priority; see Anthropic's crawler guidance.
Decide separately whether you want visibility in AI Search and whether you permit training use. Review the current controls for each provider before changing robots.txt. Avoid assuming one setting governs every company's search, user-directed retrieval, and training crawler.
What to Do This Month
- Open Search Console's generative AI performance reports and note which pages receive impressions.
- Check the Search generative AI inclusion setting for your property and document the choice.
- Review your existing crawler rules against your training preference; monitor the shared-control rollout before relying on it for Google.
- Test a small set of customer questions in AI Mode and AI Overviews, recording visible links and subsequent referral traffic.
Google's official AI Search guidance continues to emphasize useful, original content and normal Search fundamentals. A new display format does not replace those foundations.