Google has published the operating rules for a program the European Commission ordered it to build: rival search engines and qualifying AI chatbots can now license anonymized data drawn from its European search index. The revised guidance, posted August 31, sets deadlines and an eligibility bar for a Digital Markets Act mandate under Article 6(11), not a voluntary Google offering. Every tier, audit, and price cap that follows exists because Brussels required it, a distinction that shapes how much leverage Google actually has over who gets in.
Licensing agreements start going out September 17. The data itself will not be downloadable until November 16, leaving roughly a two-month gap between a signed agreement and anything an applicant can actually query.
To qualify at all, a company has to satisfy the DMA’s own definition of an online search engine and aim its service at EEA users. Control by a government outside that zone disqualifies an applicant outright, as does an active EU sanctions listing. Scale matters too: the rules set a floor of 50,000 average monthly EU searchers measured across the prior year. Track record substitutes for scale only partly, since an applicant also needs either two full years already running a search product inside the EU or, lacking that history, more than 50 million euros already committed to the business as a newer entrant. Once a company signals interest, Google’s clock starts: a reply is due inside seven calendar days, no more.
Three data tiers sit beneath full access. The free tier caps out at 1,000 rows, useful only for a first look. Two paid tiers go further: a synthetic set built from as many as 10 million queries, and a metered slice equal to five percent of the complete dataset. All three open on the same November date, though only the paid five-percent cut and the full dataset trigger an outside audit before access is granted; the free rows and the synthetic set stay ungated so a prospective licensee can judge the data’s usefulness before spending on that review.
That initial review, labeled Level 1, sends an independent auditor through the applicant’s technical setup, storage systems, and internal workflows to confirm its plans for the data look credible and its privacy safeguards are properly built, not just promised. Passing that gate is not the end of scrutiny. A second, recurring check, Level 2, follows once access is live, filed first within six months and then annually, to confirm the safeguards hold up under actual use rather than existing only on paper.
Pricing itself is not Google’s call. The Commission’s rules cap what can be charged at the incremental cost of supplying the data plus a defined return, measured against a fair, reasonable, and non-discriminatory standard. That formula removes price as a lever Google could otherwise use to slow adoption by a well-funded rival.
One requirement outlasts the paperwork. Google must keep a public page naming every outside search engine that has taken up access, a registry the Commission intends to revisit every two years. That list, once populated, becomes the clearest signal of whether the qualifying chatbots covered by the order actually show up as licensees, or whether only established search rivals bother.
Search Engine Journal first covered the underlying order in July, when the thresholds and the broad calendar were known but the mechanics were not. This update from the same outlet fills in what that earlier piece could not: the two-tier audit structure, the capped fee formula, and the specific dates data actually changes hands.
Any team competing against a well-capitalized AI answer engine has a concrete date to watch: November 16, the day the first data becomes downloadable. If a chatbot rival’s name lands on Google’s public registry in the months after, that competitor is training on the same click, query, and impression signals Google uses internally, a materially different baseline than exists today.
Search Engine Journal’s Matt G. Southern reported the updated licensing documentation on September 1, 2026, citing Google’s revised European Search Dataset Licensing Program page.
