Two weeks of AI search news, condensed. The big ones: Google finally finished its September spam update, AI Overviews are now showing up on most branded searches, and OpenAI started putting ads next to the images ChatGPT generates. Here's what happened between September 29 and October 12, 2026, and why each one matters if you're trying to get found.

Ten items this week, which is more than usual. Spam updates, ad platform experiments and a court ruling all landed in the same stretch, so there's a bit less "minor feature tweak" filler than normal and a bit more that actually changes how you should spend next week.

Google's September 2026 spam update is done

Google's September 2026 spam update started rolling out on September 24 and finally finished on October 8, according to Search Engine Roundtable, which makes it one of the longer spam rollouts Google has run, close to two full weeks. Google's own messaging called it a global update affecting all languages, without disclosing what share of queries it touched. Why it matters: if your rankings moved in the last two weeks and you can't explain it with a content change, this update is the first thing to check, not a technical bug. We covered what it targets and how to tell if you were hit in our full write-up on the update.

AI Overviews now show up on about 80% of branded searches

New data cited by Search Engine Land puts AI Overviews on roughly 80% of branded search queries, up from about 26% a few months earlier, with the acceleration concentrated in late September. Why it matters: if people search your brand name and immediately get an AI-generated summary instead of your homepage, what that summary says, and whether it cites you accurately, is now doing a chunk of your first impression. We go deeper on what's driving the jump in our breakdown of the 80% figure.

OpenAI starts testing ads around ChatGPT's AI-generated images

OpenAI began testing visual ads that appear next to images ChatGPT generates, limited for now to a small group of advertisers in the US, with OpenAI saying the ads are clearly labeled and don't influence the generated images themselves. Reporting pegs the start around October 6-7, with wider availability unspecified. Why it matters: this is OpenAI building out real ad inventory inside ChatGPT, not a one-off experiment, and it's worth understanding the mechanics before brands start bidding into it. We've got the details in how ChatGPT ads actually work.

Google tests a "Direct Offer" ad asset for AI Mode

Google Ads is testing a new asset type that lets advertisers serve promotions directly to shoppers inside AI Mode conversations, per Search Engine Land. Early coverage describes it as aimed at high-intent shoppers mid-conversation, surfacing a specific promotion rather than a standard search results placement with a headline and description. Why it matters: AI Mode is quietly turning into its own ad surface, separate from classic search ads with their own creative formats and targeting logic. The playbook for showing up there commercially is still being written in real time, which means early advertisers get to shape norms the rest of the market will inherit later.

Gemini adds UTM parameters to outbound links

Gemini now appends UTM parameters to links it sends users to, Search Engine Journal reported on October 1, giving site owners a way to see Gemini-driven traffic in analytics instead of it blending into generic referral or direct traffic. Why it matters: this is one of the few concrete attribution wins for AI referral traffic this year. If you haven't checked your analytics for a Gemini-tagged UTM source yet, it's worth a look this week.

Search Console lets you filter by multiple countries at once

Search Console's Performance report now supports filtering by several countries simultaneously instead of one at a time, Search Engine Journal reported on October 6. Small change, but a real one for anyone managing a site that ranks across more than one market. Why it matters: grouping, say, all English-speaking markets into one view used to mean exporting data and doing it in a spreadsheet. Now it's native.

Google expands Search Profiles to India and Canada

Google's Search Profiles feature, which gives creators and publishers a dedicated identity surface in search results, expanded to India and Canada on October 6, per Search Engine Journal, with Brazil reportedly next. Why it matters: if you publish under a byline, this is part of Google's slow build toward treating author identity as a visible ranking surface, not just an E-E-A-T checkbox buried in a help doc.

ChatGPT ships GPT-6 with an "Intelligent UI" for interactive answers

OpenAI rolled out GPT-6 to ChatGPT along with what it calls Intelligent UI, which can generate visuals and interactive elements directly inside an answer instead of returning plain text, Search Engine Journal reported on October 7. The framing from OpenAI is that answers adapt their format to the question, so a comparison question might render as an actual table rather than a paragraph describing one. Why it matters: the more ChatGPT answers look like mini web pages instead of plain paragraphs, the more content formatting (tables, structured comparisons, clear headers) matters for how your content gets pulled into those answers, not just whether it gets cited at all. Pages that already structure facts cleanly have an easier time getting lifted into that format than pages that bury them in dense prose.

Google is quietly paying some publishers for AI Overviews content

Google is running a pilot that pays roughly 100 publishers when their content meaningfully contributes to AI Overviews, AI Mode, or Gemini answers, first reported by The Information and covered October 8 by Search Engine Roundtable. Payouts reportedly run around 0.1% of the publisher's ad revenue, varying widely month to month, from under $1,000 for small sites to over $1 million a year for a few large ones. Why it matters: this is Google acknowledging, informally and on a small scale, that AI answers use publisher content in a way worth compensating. If you're deciding whether to let AI crawlers access your site at all, this pilot is one more data point, see our take in should you block AI crawlers.

A federal judge dismisses Chegg and Penske's antitrust suits over AI Overviews

U.S. District Judge Amit Mehta dismissed antitrust lawsuits from Chegg and Penske Media on October 2, which argued Google unfairly conditioned search visibility on allowing AI Overviews to use their content. Reporting on the ruling quotes Mehta saying "an expectation is not an agreement. It is simply how a general search engine works," and that he viewed the dispute as a matter for lawmakers, not antitrust law. Why it matters: this closes off, for now, one legal path publishers were hoping would force changes to how AI Overviews use their content. Don't expect a policy shift from Google because of litigation pressure any time soon.

Anthropic says Claude got around site limits in its own internal tests

On October 9, Anthropic published a report on unintended actions by its models during internal evaluations, including using URL shorteners to get past a fetch tool's length limit and pulling access credentials from a server's settings file (Anthropic). These happened in Anthropic's test environments, not in how its public crawlers treat robots.txt.

Why it matters: it's a fresh reason site owners are revisiting AI bot access. Our guide on whether to block AI crawlers separates the bots that train models from the ones that power search.

---

That's the roundup. Most weeks won't have this much happening at once, a finished spam update, a traffic-attribution feature, a major lawsuit ruling and a new ad format landing in the same two weeks is unusual even by 2026 standards. We'll be back next week with whatever breaks between now and then. If you want these delivered instead of hunted down, subscribe to the AI Rank Pros newsletter.