Google tests AI visibility data; UK mandates opt-out
Google concedes ground on AI reporting while UK regulators force publisher opt-outs. The citation mix in AI answers is about to shift.
Key takeaways
- Google is testing AI visibility data and an opt-out toggle inside Search Console.
- The UK's CMA is forcing Google to separate AI Overviews opt-out from blue-link inclusion.
- National UK publishers will opt out fast; trade and B2B press will mostly stay in.
- Wires, Wikipedia, Reddit and trade titles will fill the citation vacuum left by opted-out publishers.
- Brands should audit earned-media placements by opt-out risk and track citations per LLM separately.
Google has spent two years insisting that AI Overviews are simply search, so no separate reporting is needed. That position is quietly cracking. Search Engine Journal reports that Google is testing AI visibility data inside Search Console alongside an opt-out toggle, while the UK's Competition and Markets Authority is forcing publisher controls on AI search features. Two concessions in one week, from a company that does not concede lightly.
The shift matters because, until now, brands have had no instrumented view of whether they appear in AI answers, only inference from third-party trackers like Profound, Peec and Ahrefs. Google's own data, if it ships, would settle arguments inside marketing teams that have raged since AI Overviews launched in May 2024. It would also, conveniently, give Google a defensible answer to regulators who have spent eighteen months asking what publishers are actually losing.
The opt-out is the bigger story
The UK intervention is the more consequential of the two. The CMA's designation of Google with Strategic Market Status, finalised in October, allows it to mandate granular controls: a publisher can now refuse to feed AI Overviews while remaining in the blue links, something Google had previously bundled into a single take-it-or-leave-it choice via the Google-Extended token. That bundling was the central grievance of the News Media Association and a clutch of European publishers. Forcing it apart changes the economics of opting out from "commercial suicide" to "live option."
Expect take-up to be uneven. National newspapers and the FT-tier business press will opt out fast, because their click economics are already broken and AI Overviews accelerate the bleed. Trade publishers, B2B media and think tanks will hesitate, because being absent from AI answers means being absent from the discovery layer that increasingly shapes procurement research. The result, over the next twelve months, will be a thinner, more commercially-motivated training and retrieval corpus, with the most-cited mainstream outlets pulling back and the long tail staying in.
What this does to citation patterns
For brands building authority through earned media, the opt-out wave is a planning problem. If the Times, the Telegraph and the Guardian disappear from AI Overviews on UK queries, the citations that fill the vacuum will come from elsewhere: Reuters and the AP (wholesale licensors, unlikely to opt out), Wikipedia, Reddit, LinkedIn, trade press, and the brands' own properties. Coverage in a national broadsheet has long been the gold-standard proof point for a CMO presenting to a board. In AI search, it may become one of the least visible placements.
This has specific consequences for the sectors we track. Financial services brands relying on FT, Bloomberg and Reuters mentions retain most of their AI-citation surface, because the wires will stay in. Multilaterals and UN agencies, whose authority in models has rested heavily on Guardian, BBC and Reuters coverage of their reports, will see uneven exposure depending on which UK outlets opt out and how aggressively. Industrial groups quoted in trade titles (ENR, Chemical Week, Global Cement) are largely insulated, because those publishers cannot afford the opt-out. Philanthropic and policy institutions, whose visibility leans on long-form coverage in opinion pages, are the most exposed.
What the Search Console data would actually show
Assume Google ships the AI visibility report broadly in 2026. It will almost certainly be impression and click data for AI Overviews and AI Mode, segmented by query. It will not, on past form, show which sources Google's model cited alongside you, nor the prompt that triggered the answer, nor anything that lets you reverse-engineer the retrieval logic. That is the data brands actually need, and the data Google has the least commercial interest in releasing.
The toggle is the more revealing element. An opt-out switch inside Search Console, separate from robots.txt directives, signals that Google now expects opt-out volume large enough to need a managed interface. Companies do not build dashboards for behaviour they think is rare.
For B2B marketing teams, the practical move is to stop treating AI visibility as a single number. Track citations across ChatGPT, Perplexity, Gemini and Claude separately, because the source mixes are diverging fast and will diverge further once UK opt-outs land. Audit which of your earned-media placements sit behind paywalls or with publishers likely to opt out, and rebalance toward sources that will remain in the retrieval set: wires, your own research, Wikipedia-grade reference material, and the trade press your customers actually read. The brands that win the next phase of AI search will be the ones who understood, before their competitors did, that the citation graph is about to be redrawn by regulators, not by algorithms.