AI Search News June 2026: Opt-Outs, ChatGPT Ads & the GEO Data Boom

Key takeaways

Google opened a way for publishers to opt out of AI Overviews and AI Mode specifically (announced June 4) — separate from blocking Google Search entirely.
Google confirmed AI Overviews and AI Mode are finally coming to France, one of the last major holdout markets, before September 23.
A revised academic study confirmed AI Overviews cut outbound organic clicks by 39.8%, in the first randomized field experiment on the question — sponsored and internal Google clicks stayed flat.
Microsoft shipped its own AI-answer opt-out for Bing and took Copilot Cowork to general availability, making June the month both of the two largest search platforms built a way to turn AI answers off.
ChatGPT ads expanded to Japan and South Korea on June 22, joining the US, Canada, Australia, New Zealand, and a UK self-serve beta that opened June 6 — OpenAI's ad business is now live in five countries less than five months after launch.
GPT-5.5 Instant got a dedicated update for shopping and local-business queries on June 24, alongside continued rollout of product feed ads for retailers.
Perplexity spent June building agent infrastructure: a hybrid local-cloud inference system (unveiled at Computex, June 2), the SPACE sandboxed runtime for long-running agents (rollout began in June), a Microsoft 365 integration for Perplexity Computer, and enterprise deals with Docusign (June 24) and OpticOdds (June 4).
Semrush's AI Visibility Index went from a 2,500-prompt pilot to a 126-million-prompt flagship study (published June 26) — finding only 36 brands held top-100 visibility across every major AI platform every month, and that mention/citation overlap on Gemini can be as low as 30%.
Adobe formally entered AI visibility tracking with the June 17 launch of "Adobe Brand Visibility," and two newer entrants, Reputation and Citelens, launched their own GEO tools the same week.

June 2026 didn't have a single defining headline the way some months in AI search have — no billion-user milestone, no valuation record. Instead it was the month the infrastructure around AI search kept hardening: Google and Microsoft both formalized ways for people (and publishers) to turn AI answers off, an independent study nailed down exactly how much organic traffic AI Overviews take, OpenAI's ad business went from a US experiment to a five-country product, and the tools brands use to measure their own AI search visibility got both more crowded and dramatically better resourced.

Here's each story in turn.

Google opens a way for publishers to opt out of AI Overviews and AI Mode

The most consequential June move for publishers: Google confirmed on June 4 that it was testing new controls letting site owners opt out of AI Overviews and AI Mode specifically, without opting out of Google Search results altogether. The distinction matters — until now, publishers could only block Google's generative features by blocking crawling entirely (via Google-Extended), which also hurt their conventional rankings. The new domain-level control doesn't affect ranking or stop model training, and changes take effect within one to two days. Google formalized this into a Search Console toggle in the weeks that followed.

The move came the same month as a routine but relevant reminder that ranking fundamentals still matter: Google's June 2026 spam update rolled out June 24–26, its latest algorithmic pass targeting low-quality and manipulative content — rankings that feed both classic blue-link results and what AI Overviews pulls from.

France will finally get AI Overviews and AI Mode

Google confirmed on June 29, in a letter to French publishers, that AI Overviews and AI Mode will launch in France before September 23 — one of the last major markets still waiting for the features, after most of the rest of the world got them throughout 2025.

A revised study confirms AI Overviews cut publisher clicks by 39.8%

The most important AI-search research of the month wasn't new — it was a revision. A randomized field experiment involving 1,065 desktop Chrome users, authored by researchers at the Indian School of Business and Carnegie Mellon University's Heinz College, was revised and republished on SSRN on June 17. Its finding: when an AI Overview appears, outbound organic clicks fall by 39.8% and zero-click searches rise by 34.5%. Clicks on sponsored results and Google's own properties didn't move in any statistically meaningful way — meaning AI Overviews substitute specifically for organic website traffic, not for ad clicks or engagement with Google's ecosystem generally. It's the first randomized (rather than observational) study to isolate that effect, adding real methodological weight to earlier observational findings, like Ahrefs' widely cited April 2025 analysis of 300,000 keywords showing a 34.5% organic click decline.

Microsoft ships its own AI-answer opt-out, and takes Copilot Cowork to general availability

Google wasn't the only platform rethinking how much control to hand back to users and publishers this month. Microsoft made two moves that echo the Google story from opposite ends of the product.

On June 16, Copilot Cowork reached general availability inside Microsoft 365 Copilot — a distinct mode from Copilot Chat, built for longer, multi-step, autonomous work rather than single-turn question answering. It's Microsoft's clearest agentic-search-adjacent product push of the month, aimed at closing the gap with OpenAI's and Perplexity's own agent products.

Separately, and later in the month, Microsoft's search lead Jordi Ribas unveiled a Bing browser extension that lets users disable Copilot's AI answers entirely and revert to classic blue-link results — effectively a "kill switch" for AI Overviews' Bing equivalent. Coming in the same month as Google's own publisher-facing opt-out toggle, it's a notable signal: two of the largest search platforms independently decided that giving people (in Microsoft's case, end users; in Google's, publishers) an explicit way to turn AI answers off was worth shipping in the same four-week window.

ChatGPT ads expand to Japan and South Korea

OpenAI used June to scale the ad business it opened to self-serve buyers on May 5. On June 6, OpenAI's VP of Monetization confirmed ChatGPT Ads Manager opened to UK businesses in self-serve beta. By June 22, ads were confirmed live in Japan and South Korea, alongside the expanded UK self-serve rollout — bringing ChatGPT's ad footprint to five countries (US, Canada, Australia, New Zealand, UK) plus the two new Asian markets, less than five months after ads first launched in February.

GPT-5.5 Instant gets a dedicated shopping and local-business update

Shopping got dedicated model attention too. OpenAI shipped a GPT-5.5 Instant update on June 24 explicitly aimed at making shopping and local-business queries "feel more useful," and continued rolling out product feed ads to retailers built on the Agentic Commerce Protocol (ACP) — the plumbing that lets merchants push structured product data directly into ChatGPT's shopping results rather than relying on scraped or indexed content.

For brands, this is the throughline worth tracking: OpenAI is building a genuinely separate commerce and advertising layer inside ChatGPT, distinct from organic visibility in chat answers. Ranking organically in ChatGPT and buying placement in ChatGPT are becoming two different disciplines, the way SEO and paid search diverged decades ago.

Perplexity spends June building agent infrastructure

Perplexity's June was almost entirely about infrastructure and enterprise distribution rather than consumer-facing features.

At Computex on June 2, Perplexity unveiled a hybrid local-cloud inference system, splitting workloads between on-device and cloud processing. On June 4, sports-data provider OpticOdds began powering betting-market queries inside Perplexity Computer. Later in the month, Docusign's Intelligent Agreement Management platform integrated with Perplexity Computer and Computer for Counsel (June 24), letting legal teams automate contract workflows through Perplexity's agent. Perplexity also confirmed that Computer now works inside Microsoft 365 — drafting documents in Word, analyzing spreadsheets in Excel, and pulling context from Outlook and Teams.

Underlying all of it: Perplexity began rolling out SPACE, a sandboxed runtime for long-running AI agents, as the infrastructure layer for Computer sessions in June (the company detailed the system publicly in mid-July). SPACE isolates each agent task in a disposable virtual machine, supports pausing and resuming sessions across hours or days, and cut median sandbox creation time from 185ms to 60ms compared to Perplexity's prior system — the kind of unglamorous plumbing work that matters more for Perplexity's stated ambition of becoming an "agent company" than any single consumer feature would.

None of this shows up in Perplexity citation tracking the way a search feature launch would, but it's a clear signal of where Perplexity is placing its bets.

Semrush's AI Visibility Index scales to 126 million prompts

June was the month AI-visibility measurement grew up. On June 26, Semrush published its expanded 2026 AI Visibility Index, scaling its flagship study from an initial 2,500 prompts to 126 million real AI search prompts across major platforms. The findings are a useful reality check for anyone assuming AI visibility is evenly distributed:

Only 36 global brands — the "Universal 36," including YouTube, Google, Reddit, Amazon, Facebook, Apple, Walmart, Disney, and Nintendo — maintained top-100 visibility across every tracked platform, every month of the study.
Mentions and citations aren't the same thing. On Gemini, the overlap between brands mentioned in an answer and domains actually cited as sources can be as low as 30% — meaning a brand can be talked about without being credited as the source, or vice versa.
Visibility concentration varies sharply by category. In News and Media, the top three brands captured 82.9% of total category visibility; in Consumer Electronics, 76.9%. Finance (41.4%) and Industrial (42.2%) were far more open, suggesting more room for smaller brands to break through in those categories.
45% of marketing leaders say they can't accurately measure their brand's AI visibility, and only 9% say they have tools to track all the relevant metrics across platforms — the gap this entire tooling category exists to close.

A separate, narrower study reinforced the same theme from the content side: Notified's analysis of over 8,000 GlobeNewswire-distributed press releases, also published in June, found 99.3% were cited by ChatGPT or Claude within hours of publication — average time to first AI citation was just 8 hours — and that well-structured releases generated 2.6 times more citations on average than poorly structured ones.

Adobe formally enters AI visibility tracking, and new GEO challengers launch

The category of tools — including Airefs — built to track how brands show up in AI-generated answers got more crowded in June, and one of martech's biggest names planted its flag in it directly.

On June 17, Adobe launched "Adobe Brand Visibility," a unified product combining Semrush's AI-visibility tracking data with Adobe Experience Cloud, following Adobe's completed acquisition of Semrush in April. Coming the same week as Semrush's own expanded AI Visibility Index, it's the clearest signal yet that AI search visibility tracking has gone from a niche startup category to something the largest martech vendors want to own outright.

Smaller entrants moved the same direction. Reputation launched a GEO Readiness Audit on June 24, aimed at multi-location brands trying to baseline how AI engines represent their business listings. A day later, on June 25, a new platform called Citelens launched, positioning itself specifically around generative-engine visibility analytics for corporate profiles.

What this means if you're tracking AI search visibility

Taken together, June's news points in one direction: AI search is maturing from "a new surface to watch" into infrastructure with real commercial mechanics — opt-out controls on two major platforms, paid placements, agent runtimes, and a genuinely competitive, increasingly data-rich tooling market. A few practical implications:

If you're a publisher, Google's new opt-out toggle (and Bing's parallel move) are worth evaluating carefully — they let you exit AI answer surfaces without sacrificing your regular rankings, which wasn't previously possible on either platform.
If you sell products, ChatGPT's ad and shopping infrastructure is no longer experimental. Structured product feeds via ACP and ChatGPT Ads Manager are now live in five countries — worth testing well before your competitors do.
Semrush's finding that only 36 brands hold consistent cross-platform visibility, and that mention/citation overlap can be as low as 30% on a single platform, is a strong argument against assuming visibility on one AI surface implies visibility on others. Coverage breadth is still the first thing to check in any AI visibility tool comparison.