Week in Brief
- This was a busier week than the algorithm trackers alone suggest. The story was not a core update. It was Google formalising who gets to appear in AI answers, and who can leave.
- Google documented a Search Console setting that pulls a site out of AI Overviews, AI Mode, and Discover’s generative features.
- Google switched on AI Overviews and AI Mode in France, its last major market.
- ChatGPT began testing new ads from different advertisers in answer footers
Search Console documentation for a control that removes a site from AI Overviews and AI Mode, rolling out beyond the UK
Google published Search Console help documentation for a domain-level setting called “Search generative AI Control”. It began rolling the control out to a subset of site owners beyond the original UK group.
The setting gives website owners a binary choice: remain eligible to appear in AI Overviews, AI Mode, and generative features in Discover, or opt out of all these experiences.
Google’s documentation states that opting out removes impressions and traffic from those features but does not affect standard organic rankings. The control first appeared in the UK on 3 June under a binding UK CMA order; we broke down what that launch meant for B2B organic strategy in our Weekly edition on the Search Console AI opt-out control. This week is its documented expansion.
What this AI documentation means for B2B, ITES, and SaaS
This forces a decision most teams have been able to defer. If your top informational pages (“what is GEO”, “how does LinkedIn Ads work”) are being summarised in AI Overviews without earning the click, the control now lets you pull them from those surfaces without touching your blue-link rankings.
However, for most companies investing in GEO, opting out should be considered a tactical decision. Leaving these features would also remove the possibility of your website being cited as a source within AI-generated answers.
ChatGPT begins testing new ads from different advertisers in answer footers
ChatGPT is testing two ads from two different advertisers in the footer of its answers. This is a test, not a confirmed rollout, and OpenAI has not published ad-targeting or placement documentation. It is the clearest signal yet that paid placement is moving toward the answer itself, not only the sources around it.
What this new ads format means for B2B, ITES, and SaaS
For B2B companies that have treated ChatGPT as an organic citation channel, this is the sign the economics may shift toward paid. Nothing changes for your GEO work this week, because the test is limited and unpriced. The teams to watch it hardest are those already running paid search, because a ChatGPT ad unit would compete for the same bottom-funnel intent. Keep building citation visibility, and wait till ChatGPT makes it official.
On Our Radar for Next Week
A new critical survey of 45 GEO studies published between November 2023 and July 2026 is likely to sharpen the debate around how AI visibility is measured and marketed. Its central conclusion is not that GEO is ineffective, but that the evidence is far narrower than many commercial claims suggest.
Within the reviewed research, no technique demonstrated a stable, longitudinal and cross-platform causal improvement in organic AI discoverability, clicks or conversions. Most successful experiments showed that content could influence an answer after it had already been retrieved, which is materially different from proving that an AI engine is more likely to discover, retrieve, or recommend the page in the first place.
Expect greater scrutiny of GEO platforms and agencies that report a single “visibility score” without separating search activation, retrieval, citation, prominence, factual use, and downstream traffic. The more defensible approach is to measure each stage independently, repeat tests across engines, prompts and dates, and treat visibility as a changing distribution rather than a permanent AI ranking. The practical takeaway remains constructive: create relevant, verifiable, clearly structured and technically retrievable content, but be cautious of universal formulas and percentage-lift promises that are not supported by end-to-end evidence.
White Bunnie Observation
Based on our review of multiple Google Search Console properties, the Search generative AI control is not yet appearing universally. So far, we have observed the setting only on properties that have also received access to the dedicated Generative AI performance report.
This suggests that Google may currently be enabling the reporting and control features for the same group of eligible properties, or rolling them out in closely connected phases. However, Google has not officially confirmed that access to the performance report is a mandatory prerequisite for receiving the control.
The absence of this setting does not necessarily indicate a technical problem with the website or its Search Console configuration. It may simply mean that the property has not yet been included in Google’s phased rollout. Google’s published documentation currently confirms that the control is being made available to only a subset of website owners.
Source: Screenshot captured by White Bunnie from an active Google Search Console property. The property name has been redacted to protect client confidentiality.
Questions We Were Asked This Week
Now that we can, should we opt our blog out of AI Overviews?
For most B2B and SaaS companies, no. The new Search Console control removes your pages from AI Overviews, AI Mode, and Discover’s generative features. Google says that does not affect your standard rankings. But leaving those surfaces also removes you as a citable source inside them. If you are building GEO visibility, opting out trades away the citation you are working to earn. The move suits click-dependent publishers far more than pipeline-driven companies.
We provide services into France. Did anything actually change for us this week?
Yes. Google switched on AI Overviews and AI Mode in France on 22 July, the last major market to get them. If your French-language pages answer general category questions, expect the same summary-above-organic pattern as visible in other countries since 2024. Your AI-impression baseline in France starts at zero this week, so a year-on-year traffic comparison that crosses 22 July will contain a structural break.
Should we start budgeting for ChatGPT ads?
ChatGPT is testing new variants of the ads with different advertisers in answer footers, but it appears to be a limited test with no published targeting or pricing. Keep investing in citation visibility, which is what earns you presence in the answer today. Watch OpenAI’s official channels for a real ad product before you move any spend.
