The EU's Google Decision Is In, and AI Chatbots Made the List
Back in June I wrote about a European Commission deadline that almost nobody in marketing was watching, and I told you the single line to look for was whether AI chatbots survived on the eligibility list for Google's search data. The Commission didn't wait for its own deadline. On July 16, eleven days ahead of the statutory date, it adopted two binding decisions against Google under the Digital Markets Act (European Commission), and the line I told you to watch came back the way the AI companies wanted. Chatbots with search functionality are on the list (Commission developer portal).
That answers the question my June post could only reason around, and it starts a clock. Google's search data begins flowing to eligible competitors in January 2027, and the Android side of the ruling opens up in July 2027 (Digital Watch). Since I asked you to watch this one with me, this post is the follow-through: what the Commission adopted, the clauses that will decide how much it matters, and what I'd do with the runway between now and January.
What the Commission adopted
There are two decisions, and the device one deserves as much of your attention as the data one.
The first covers search data under Article 6(11). Google has to give eligible third parties access to anonymized ranking, query, click, and view data on fair, reasonable, and non-discriminatory terms, with the decision spelling out a pricing formula the Commission calls fair (MediaPost). Eligible parties include rival search engines and, the part that matters for this blog, AI chatbots that offer search functionality. That's the OpenAI and Anthropic category, confirmed in the Commission's own materials rather than inferred from them.
The second decision covers Android under Article 6(7). Rival AI assistants get equally effective access to eleven system features that Gemini uses today, including the kind of hotword voice activation that "Hey Google" enjoys, so a European could set a competing assistant as the one their phone wakes up for (TechTimes, MediaPost).
What makes these decisions different from the DMA's earlier, vaguer rounds is the mechanism. These are Article 8(2) specification decisions, and the format is a build spec rather than a statement of principles: named data fields, defined anonymization methods, the pricing formula, eligibility criteria with thresholds, and audit standards, all binding without waiting for any finding that Google broke the rules (TechTimes). Non-compliance exposure runs up to 10 percent of Alphabet's global turnover, a number north of 35 billion dollars, and while an appeal is widely expected, an appeal doesn't pause the obligations. Executive Vice-President Henna Virkkunen pitched the upside plainly: "All developers, large and small, are welcome to explore these new opportunities" (Commission press release).
The clauses that YOU MIGHT HAVE MISSED
I spent time in the decision materials so you don't have to. Google gets a risk-review lever. Before sharing data with any specific third party, Google may assess whether that particular recipient poses serious cybersecurity or data-protection risks (MediaPost). A reasonable safeguard on its face, and also an obvious place for friction, since the company writing the risk assessments is the company that argued against sharing at all. Kent Walker, Google's president of global affairs, made the company's position clear after the ruling: "Today's decisions risk undermining vital privacy and security guardrails for millions of Europeans" (Brussels Signal). Expect that argument to resurface recipient by recipient.
The anonymization is layered, and the Commission reserved the right to change it. The method was developed with privacy experts and tracks the draft joint DMA-GDPR guidelines with the European Data Protection Board, and the decision says the Commission may amend the measures based on how the market develops, including independent third-party evaluation (Commission press release). Read that as a spec that can move under your feet, in either direction, after January.
The whole thing, meanwhile, is binding without being a verdict. Neither decision finds that Google violated the DMA, and neither imposes a fine today; they define what compliance looks like, subject to judicial review (Digital Watch). Not everyone thinks the design works even on its own terms. Dirk Auer at the International Center for Law & Economics expects the ruling to shrink the number of AI assistants available to Europeans rather than grow it, on the theory that the privacy and compliance load falls hardest on smaller players (Brussels Signal).
Updating the June framework
My June post sketched a two-track world for AEO, and I labeled the European track a hypothesis. The hypothesis now has dates and named beneficiaries, so the sketch deserves an update.
In the EU, starting in January, the AI engines your European buyers use can learn from the same ranking and click signals that made Google Search good. Whether that re-couples your Google rankings to your AI citations is still my open question from June, and I'll hold my answer until there's data, since aggregated, anonymized signals improve a model's sense of relevance in general without handing it a list of pages to cite. My lean is unchanged, that high-commercial queries in Europe are where any re-coupling shows up first, and it stays a lean rather than a finding.
Two things the decision does not do deserve equal billing. It grants no access to Google's index and nothing about your individual pages, so this is signal-sharing rather than syndication. It also changes nothing in the United States, where the decoupling I've been documenting all year keeps running on its own physics.
The Android half is the sleeper for marketers, and I'd argue it's underrated in the coverage. Citation strategy in Europe currently assumes Gemini sits on every Android phone as the default. From July 2027, a European can wake a rival assistant by voice, so assistant share on the continent becomes contestable in a way it hasn't been, and the question of which engine's citations matter in Europe gets a mechanism for changing.
WHAT TO DO
Baseline your EU citations now. January gives every brand selling into Europe a natural experiment, a clean before-and-after window on whether shared search signals move AI answers, and the experiment only works if you captured the before. Split your AI-visibility tracking by region this quarter, log which engines cite you for your money queries in Europe versus the U.S., and date-stamp it.
Keep the SEO fundamentals in shape, weighted toward Europe. The scenario where rankings pull double duty, feeding classic search and the AI engines reading Google's signals, now has a start date instead of a maybe. Nothing about that argues for new spending so much as it argues against letting the boring work slide this year.
Watch who clears the gate, because the eligibility list on paper and the beneficiary list in practice may differ. Between the risk reviews, the pricing formula, and the appeal, it's an open question which AI providers are receiving data on day one, and the answer tells you which engines to prioritize for European visibility. I'd also keep half an eye on the amendment clause, since a spec the Commission can revise is a spec worth re-reading each quarter.
Treat all of it as calendar work rather than a fire drill, a little now so that January finds you holding a baseline instead of starting a scramble. The Commission moved eleven days early, which leaves you five months, and that's more warning than this industry usually gets.
Jarred Smith is the author of Explainable: Why AI Recommends Some Brands & Ignores Others, an Amazon bestseller on AEO, GEO, and SEO. He's a marketing leader with nearly 20 years of experience across healthcare, public media, retail, and environmental services. Find him at jarredsmith.com.