Google sues the EU to avoid sharing users' search history with rivals

Google has gone to the EU General Court in Luxembourg to overturn two European Commission decisions adopted on 16 July under the Digital Markets Act. One of them makes Google hand rival search engines the data it collects to improve its own search; the other opens Android to competing AI assistants. The company's stated reason for the appeal is privacy.

"We're appealing decisions that will force us to share people's private search history without sufficient anonymisation and weaken vital security protections on Android", said Oliver Bethell, Google's senior director for competition. "People use Search for their most personal questions, from medical worries to close relationships, and mandating we share these personal queries without adequate safeguards would cause irreversible harm to user privacy."

The obligation itself comes from Article 6(11) of the DMA, and the Commission's decision spells out how it should work. Google must start sharing search data with eligible search providers from January 2027, and the Android changes are due in July 2027. Eligible recipients go beyond classic search engines: the Commission says AI chatbots with search functions can receive the data too.

What the Commission says about anonymity

The Commission's own page is specific on the point Google disputes. Its decision sets out "a multi-layered method to anonymise the shared data, developed in close collaboration with internal and external privacy experts", aligned with the draft joint guidelines on how the DMA and the GDPR interact, written with the European Data Protection Board. The data to be shared is described as "the same data that it collects to optimise its own search services".

So the two sides are not arguing about whether search queries are sensitive. They agree on that. They are arguing about whether anonymisation can hold once the logs leave the company that produced them, and that argument has a history worth remembering.

Why "anonymised queries" is a loaded phrase

In August 2006 AOL published about 20 million search queries from 650,000 users, with usernames and IP addresses stripped out, for research purposes. Within days journalists had matched user number 4417749 to a named 62-year-old woman in Georgia purely from what she had typed. The Electronic Frontier Foundation, which filed a complaint over the release, has kept the case as the standard illustration that a query log is a fingerprint even with the identifiers removed.

That does not make the Commission wrong. Twenty years on, anonymisation is a different craft, and the decision was written with privacy regulators rather than around them. It does mean Google's objection is not obviously a pretext, and that the interesting detail is technical rather than political: exactly how much is aggregated, how rare queries are handled, and who audits the result.

There is also an awkward asymmetry in Google's position. The queries in question are already collected, stored and used by Google itself, and the appeal is about who else gets them rather than whether they exist. Users concerned about "the most personal questions" have the same exposure today, just to one company instead of several, which is the part a competition regulator is least equipped to fix.

The practical reading for anyone in the EU: nothing changes in your search box this year, and the January 2027 date applies to data sharing between companies, not to a new consent screen. If your search history matters to you, the lever you actually control is what you let a search engine keep. That is the same lever as when shared AI chats turned up in Google's index, and when contractors turned out to be reading Copilot prompts and looking at uploaded photos: the data you hand over is the data that can travel.