Qualcomm has worked with AI firm Trustlook to develop the SECURE AI MP Token to run alongside Qualcomm’s Haven security platform, which will debut on the Snapdragon 845 flagship chipset. The powerful silicon features a more advanced version of the machine learning platform that debuted in the Snapdragon 835, and that’s what will power the AI-based SECURE AI MP Token and its accompanying engine within the Haven platform. The token integrates with the platform to bring enhanced security through device attestation by working at the hardware level. Authentication is done through communication between the token on the hardware side and a corresponding software token. The token is generated and verified by on-board AI, with an engine and model that Trustlook built specifically to run on the Snapdragon 845.
The technology can not only be used for boot and code verification but also fraud prevention. In a demo shown to journalists at a live event, a Snapdragon 845-powered test device was able to see through a location spoofing attempt by verifying the code’s origin. Upon finding that the instructions to change the device’s location data did not come from the proper place, being the device’s location service layer and the hardware that drives it, the platform was able to automatically label the forged location data as fraudulent, and then obtain and assert proper location data to ensure that no hidden code changed the location data in spite of the existing security system. It also notified the user that suspicious activity had been detected.
Qualcomm Haven and Trustlook’s SECURE AI MP Token will debut on the Snapdragon 845 but will likely land on other modern Snapdragon chipsets going forward through the use of an AI coprocessor. By verifying code and instructions through the hardware and software, the platform will make it much harder for hackers to find exploits that can actually get past a device’s security measures. This kind of dual-layer security is just one possible use of the powerful AI systems found inside modern mobile chipsets. Undoubtedly, more use cases, user-facing and otherwise, will continue to pop up as AI becomes more commonplace in the mobile world and the ecosystems behind the chipsets powering it mature over time.