Designed to flag suspicious behavior before a transaction takes place, the system can spot tactics such as impersonating public officials, financial institutions, or acquaintances, as well as attempts to coax transfers or siphon personal credentials. Crucially, the AI model runs locally on the handset rather than transmitting conversation records to external servers, protecting user privacy while monitoring for fraud.
The tool is currently geared toward Android devices, though deployment channels and a formal rollout schedule have not been made public. For the expatriate and Chinese business communities across Southeast Asia who rely on mobile chat applications for daily commerce, local on-device screening could add a practical safety net against prevalent social engineering schemes.
Best AI model at end of October? Market says Google 77% → Like for points, predict to win points｜ToNanyang Prediction Market: https://tonanyang.cc/webapp/
