Google warns EU data-sharing plan risks privacy breaches via AI
Google warned EU regulators that a proposed data-sharing mandate could expose user privacy by enabling AI-driven re-identification of individuals from supposedly anonymised search data, demonstrating that their red team could de-anonymise users in under two hours.
What happened
Google scientist Sergei Vassilvitskii warned EU antitrust regulators that a European Commission proposal requiring Google to share search engine data with rivals such as OpenAI risked exposing users' private information.
The proposal would require Google to provide rival search engines access to ranking, query, click and view data on fair, reasonable and non-discriminatory terms. Google's AI red team demonstrated that they could re-identify users from the Commission's proposed anonymised personal data in less than two hours.
Reported impact
Google warned that the Commission's approach to anonymization may not be strong enough to prevent modern AI tools from sifting through the data to identify people, potentially harming Europeans' privacy.
Index analysis
The concern centres on whether anonymization techniques proposed by the European Commission would resist de-anonymization attacks by modern AI systems capable of re-identifying individuals from supposedly anonymised data.
Governance controls
Google proposed a broader approach with better guardrails to protect user privacy while enabling data sharing. The European Commission indicated it would finalise the proposal in coming weeks following feedback from interested parties, with a decision expected by July 27.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.