Using AI to stop tech support scams in Chrome

Google Chrome 137 introduces an on-device Gemini Nano LLM that analyzes web pages for scam indicators (e.g., pop-up warnings, full-screen takeovers, keyboard/mouse lock) and sends summarized signals to Safe Browsing—enhancing detection of fast-moving tech support scams with user-consented, privacy-preserving processing. The feature aims to improve confidence in scam verdicts, complements existing protections, and will expand to other scam types and Android, with ongoing work to address issues like prompt injection and timing bypass.