The innovations revealed in a blog post by project organisers Lyubov Farafonova and Steve Kafka [no really? ed] , are designed to safeguard users while prioritising their privacy.

Dubbed Scam Detection is supposed to tackle the alarming rise in scam calls, which cost individuals over $1 trillion annually. Using advanced on-device AI, the feature identifies suspicious conversation patterns, such as callers impersonating banks to solicit urgent fund transfers.

Farafonova and Kafka said: "Scam Detection uses powerful on-device AI to notify you of a potential scam call in real time by detecting conversation patterns commonly associated with scams."

The system provides audio and visual alerts to warn users of potential risks.

Scam Detection is off by default, with users completely controlling its activation. Notably, all processing occurs on the device, ensuring no audio or transcription is stored or shared with Google.

The feature is powered by Google's Gemini Nano AI model and is currently available for English-speaking beta users in the US with Pixel 6 or newer devices. Google plans to expand its availability in the coming months.

Meanwhile, Google is enhancing its Play Protect system with live threat detection to guard against malware and unsafe apps. This feature focuses initially on stalkerware, malicious software designed to collect sensitive data without user consent.

By analysing behavioural signals and app activity patterns, live threat detection identifies malicious apps that might otherwise evade detection by lying dormant. When a threat is found, users receive an immediate alert, allowing them to act swiftly.

All data processing occurs via Google’s Private Compute Core, which enables advanced protection without collecting or storing user data externally.

The feature is now rolling out to Pixel 6+ devices and will extend to other Android phones soon.