Google’s generative AI control turns inclusion into a property-level governance decision
Google’s detailed Search Console documentation explains how site owners can include, exclude or inherit participation in AI Overviews, AI Mode and generative AI features in Discover. The practical issue is no longer only optimization; it is governance over whether content is eligible to be linked and used for grounding.
Editor’s Note by Dageno: Dageno reorganized Search Engine Roundtable’s report and Google’s help documentation into a governance brief for GEO, SEO, publishing and analytics teams. It separates AI-search eligibility, model training controls, indexing and property inheritance so teams do not confuse one control with another.
Understand the article in 60 seconds
- The control became effective on June 17, 2026. Google says it applies to AI Overviews, AI Mode and generative AI features in Google Discover for the subset of site owners receiving access.
- Three property choices exist. A property can include content, exclude it, or inherit the nearest configured parent setting; inclusion is the default for standalone properties.
- Exclusion removes links, grounding eligibility, impressions and traffic from covered AI features. Google says the setting does not act as a ranking or inclusion signal for other parts of Search.
- This is not the same as blocking training or indexing. Google-Extended addresses model training, while noindex removes pages from Search; the generative AI control only governs specified AI surfaces.
Google has separated AI-search participation from indexing and model training
Google’s new documentation clarifies that generative AI participation is a distinct layer of content governance. A page can remain indexed in classic Search while being ineligible to appear as a link or grounding source in covered AI features. It can also be excluded from training through Google-Extended without changing this Search participation setting.
That separation matters because teams often collapse crawling, indexing, training and answer-surface eligibility into one idea. They are different controls with different business consequences. GEO leaders need to know which layer they are changing and which traffic, visibility or licensing objective the change is meant to serve.
Before changing the toggle, document the objective: AI-search visibility, model-training permission, classic indexing, or commercial licensing. One control does not solve all four.
The control has six operational rules teams must understand
Scope
The control covers AI Overviews, AI Mode and generative AI features in Discover. A single property decision can affect several answer surfaces at once.
Availability
Google began with a subset of site owners, and Search Engine Roundtable reported UK-only support at the time. Do not assume every property or market has access.
Default include
“Include my site’s links and content” is the default for properties that are not inheriting. Sites participate unless an authorized owner changes the setting.
Exclude effect
Exclusion prevents links and content from appearing or grounding responses and removes related impressions and traffic. The cost is visibility, not merely data reuse.
Propagation time
Google says exclusion generally takes a few days, with content expected to disappear within 1–2 days after the control goes live, though caches may delay some removals. Reporting should allow for propagation lag.
Inheritance
A child property inherits the nearest parent that has a manually configured value, otherwise ultimately the top-level domain. One parent change can cascade through subdomains and URL-prefix properties.
Eligibility, inheritance and propagation determine what disappears
The setting decides whether crawled content may be linked or used as input for generated Search responses. It is not a quality score.
Domain, subdomain and URL-prefix properties form an inheritance tree. A child can override the parent, but only if its owners configure it directly.
A saved choice must move through Google’s serving and caching systems. Temporary residual visibility does not necessarily mean the setting failed.
Opting out trades answer-surface presence for tighter content control
Including content preserves the possibility of links, grounding, impressions and traffic, but does not guarantee selection. Excluding content removes that possibility from the covered features. For publishers, the decision is therefore a trade between distribution and control rather than an optimization switch.
The control also changes measurement baselines. Once a property is excluded, a decline in AI impressions may be expected behavior rather than an algorithmic loss. Teams must annotate the change and avoid comparing post-exclusion visibility with an unchanged competitor set.
Governance mistakes can remove far more visibility than intended
Parent-level cascade
A domain-property exclusion can silently affect many child properties, including editorial, ecommerce and support sections.
Control confusion
Using this toggle when the real objective is training opt-out or complete deindexing creates a false sense of protection.
No baseline
Blocking before exporting AI-performance data makes it difficult to quantify the traffic and visibility sacrificed.
Fragmented ownership
Legal, editorial, SEO and product teams may change different properties without a shared policy, producing inconsistent participation.
Publishers need an approval process before changing the setting
Inventory the property tree
List domain, subdomain and URL-prefix properties and identify which parent currently controls each child.
Define the business objective
Separate Search visibility, training permissions, licensing, privacy and indexing requirements.
Export a baseline
Capture AI impressions, pages, countries, devices and dates before changing any setting.
Approve and document
Record the owner, reason, scope, expected effect and rollback condition for every override.
Validate after propagation
Check covered AI surfaces after the expected delay and confirm that classic Search visibility remains unchanged where intended.
Five teams must share ownership of the include-or-exclude decision
| Team / owner | What changes | What to do next | Metric or signal |
|---|---|---|---|
| GEO / AI visibility | Participation can be switched off at property level. | Model expected answer-share loss and identify which prompt families depend on the property. | AI impressions and prompt visibility |
| SEO / web operations | Inheritance can affect large URL groups. | Map Search Console properties and test parent-child overrides in a controlled order. | Affected URLs and classic Search stability |
| Legal / policy | The toggle governs Search AI use, not model training. | Translate licensing and consent requirements into the correct technical control. | Policy-control alignment |
| Publishing / content | Exclusion removes both links and grounding opportunities. | Decide which content portfolios require reach and which require tighter distribution control. | Portfolio-level citation and traffic value |
| Analytics / finance | The setting creates an intentional break in time series. | Annotate change dates and estimate foregone visits, leads and subscription assists. | Pre/post incremental value |
Measure the commercial value at risk before and after a control change
Record impressions by page, country, device and date before the change so lost distribution can be quantified.
Estimate which non-branded and commercial prompts rely on the excluded property as a source.
Track the percentage of previously visible URLs that disappear within the expected window.
Monitor standard impressions and clicks to verify that the AI control is not being confused with unrelated ranking changes.
Dageno’s view: AI participation is now a portfolio decision, not a robots-file shortcut
Google has made a publisher’s participation in generative Search more explicit, but the choice is not binary at the organization level. Different properties can inherit or override different settings, which turns the decision into portfolio governance.
Dageno’s editorial judgment is to treat the control as a high-impact configuration change. Establish ownership, export evidence, simulate visibility loss and document the distinction between Search AI eligibility, training and indexing before touching it.
Turn AI visibility signals into execution.
Dageno AI is a complete execution platform from data monitoring → strategy → content generation → result attribution. Monitor AI answer surfaces, identify strategy gaps, generate content actions and connect visibility changes to business impact.
Alexa for Shopping