The European Commission’s Digital Markets Act order would require Google to share Europeans’ search records with rival services after applying safeguards intended to make users indistinguishable within groups of at least 1,000. The audit tested those safeguards against real search data and found that grouping by day, area, device, and search language overlooks the language configured in a user’s browser. That omission can leave a person isolated even among tens or hundreds of thousands of records. The researcher identified roughly half a million isolated searches in several countries during August 2026; even fallback grouping at the country level left about 85,000 searches isolated. Because singling out can identify someone under data-protection standards, the audit argues the data is not anonymous.
The audit also found that supposedly separate fragments of no more than three searches could be linked into longer histories: more than 200,000 sets in Germany alone. Linked queries can reveal sensitive details, including health conditions, religion, and political views, which appeared three to five times as often as in individual fragments. Searches near military bases and diplomatic centers also suggested potential counter-intelligence risks. The researcher concludes that releasing the dataset in its proposed form would expose personal data and could amount to a large-scale breach. The findings challenge draft European privacy guidance for overlooking linkage within a single dataset, and argue that anonymity must be tested empirically, especially as AI improves inference. Google commissioned and paid for the audit; the researcher says the work remained independent.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.