TEE-based Secure and Verifiable Video Communication

The development of artificial intelligence (AI) poses a serious threat to the security of video communication systems because an adversary can generate vivid fake content that is difficult to detect. Most existing technologies for mitigating injection attacks are detection-based, that is, they use various algorithms to detect faked content. Instead of designing new detection algorithms for fake multimedia content, the proposed project will leverage trust execution environment (TEE) technology to prevent fake multimedia content from entering the communication system at the beginning. Specifically, the proposed solution builds a safe environment that includes all the essential software components for the video communication system, such as the camera driver and client software. The server maintains a database of signatures of all recognized environments, and a client is allowed to join a communication session only if its local environment can be recognized. The security of the proposed solution relies on the safety of the underlying TEE system and the associated cryptographic tools. Compared with existing technologies in the market, the proposed solution disables potential channels through which an adversary can inject its content into the communication system and does not rely on any detection methods. Therefore, it is immune to any advancement in fake multimedia content generation and provides robust yet universal protection of video communication systems. Furthermore, the proposed solution can be deployed together with existing protection mechanisms.