Scam.ai's Halo model brings local deepfake detection to Qualcomm chips
scam.ai launched halo, an on-device deepfake detection model running on qualcomm hardware at computex 2026. this partnership moves security scans to the edge, reducing latency and privacy risks for live video calls.
scam.ai unveiled halo, an on-device deepfake detection model designed to identify synthetic media in real-time during live video calls, announced at computex 2026. the launch coincides with a strategic partnership with qualcomm, integrating halo's capabilities directly into the chipset to enable local processing without cloud dependency. this architecture mitigates privacy leaks by keeping biometric data on the device and reduces the latency often associated with server-side detection. the collaboration signals a hardware-focused shift in ai safety, pushing security capabilities from centralized servers to the edge.