As physical artificial intelligence (AI) becomes capable of operating in real-world environments, ensuring cybersecurity from the vehicle design stage is becoming increasingly critical, according to Song Jong-hyuk, head of the Automotive Security Threat Research Institute at Autocrypt. Autocrypt is an AI mobility cybersecurity company established in 2019 through the spin-off of the automotive security business unit of Pentasecurity. The company provides in-vehicle software security, vehicle-to-everything (V2X) technologies, and automotive cybersecurity solutions. Physical AI refers to AI models capable of perceiving, making decisions, and controlling physical devices such as vehicles and robots. Put simply, it enables AI to move beyond software and interact directly with the physical world. As software-defined vehicles (SDVs) become central to the automotive industry and physical AI emerges as a key technology trend, automotive cybersecurity is expanding beyond vehicles themselves to encompass cloud platforms and road infrastructure. Leveraging expertise accumulated through autonomous driving security, Autocrypt is broadening its business into the physical AI sector. "Security policies must be properly incorporated from vehicle design through the development process, and we are building an architecture that integrates security technologies across the entire infrastructure, including communications between servers and roadside equipment," Song said. "The goal is to apply security throughout the entire lifecycle so that physical AI can operate safely in the real world." In physical AI environments, cyberattacks targeting sensor data or remote control commands could directly result in accidents. Song said preventing such incidents requires security specifications to be incorporated from the earliest stages of development and integrated protection across in-vehicle systems, cloud platforms, and roadside infrastructure. One of the organizations under Song's leadership is the company's Red Team, which systematically identifies vulnerabilities in automotive cybersecurity systems by examining external attack paths into vehicles. "The Red Team's highest priority is to identify external entry points that attackers could