Google scientist warns EU data plan
- Google scientist Sergei Vassilvitskii warned EU regulators on May 5 that draft Digital Markets Act search-data sharing rules could expose users. - He said Google’s red team re-identified people from the Commission’s proposed anonymised dataset in under two hours using modern tools. - The clash matters because Brussels wants to weaken Google Search’s dominance, but privacy law may limit how far data-sharing remedies can go.
Search data is the raw material here — the record of what people ask, click, and ignore. The EU wants some of that data shared with rivals so Google Search stops being such a closed fortress. But this week Google pushed back hard. Sergei Vassilvitskii, one of the company’s top privacy scientists, told EU regulators on May 5 that the Commission’s proposed anonymisation method can be reversed fast enough to put user privacy at risk. (money.usnews.com) ### What is Brussels trying to force Google to share? The European Commission’s draft remedy sits under the Digital Markets Act, the law the EU uses to make “gatekeeper” platforms open up key services. In Google’s case, the Commission proposed that third-party search engines get acce(money.usnews.com)they can build better search products and compete more seriously with Google. (digital-markets-act.ec.europa.eu) ### Why does that data matter so much? Because search quality is a feedback loop. The more queries and clicks a search engine sees, the better it can tune ranking systems, spot intent, and learn what “good” results look like. That creates a moat. Brussels(digital-markets-act.ec.europa.eu)n still point back to real people. (digital-markets-act.ec.europa.eu) ### What exactly did Google’s scientist warn? Vassilvitskii said Google’s internal red team could reconstruct identities from the Commission’s proposed anonymised dataset in less than two hours. He argued that newer AI tools make re-identification easier, (digital-markets-act.ec.europa.eu) broke it quickly.” (thenextweb.com) ### Why isn’t anonymisation enough? Because anonymisation is not magic. If a dataset keeps enough useful detail — unusual searches, timestamps, click patterns, location hints, device signals — a determined analyst can sometimes stitch those fragments back together. Think of it like a jigsaw (thenextweb.com)anonymous. That tension is the whole fight. (thenextweb.com) ### Where does GDPR come into this? If shared data can be re-identified, then it may not count as safely anonymised in the first place. That creates a collision between two EU goals that usually travel together but do not always fit neatly — stronger competition and stronger privacy. Google (thenextweb.com)gher. (money.usnews.com) ### Why mention OpenAI and other rivals? Because this is not just about classic search engines anymore. Reuters’ reporting said the proposed sharing could benefit rivals such as OpenAI, which shows how much the search market has changed. The Commission is not only thinking about ten (money.usnews.com) ### What happens next? The Commission sent Google preliminary findings and proposed measures in mid-April, and the process is moving toward a final decision by July 27. Google has said it wants to work with the Commission on stronger safeguards, but the broader fight is now out in the open. This is no longer just an antitrust remedy debate — it is a test of whether Europe can force data openness without breaking its own privacy model. (digital-markets-act.ec.europa.eu) ### Bottom line? Brussels wants to pry open Google Search’s advantage. Google wants to prove that the pry bar could crack user privacy first. If the company is right, the EU may have to choose between a weaker competition remedy and a much harder technical privacy standard.