The New Risks of Automotive AI: Uncovering LLM and MCP Security

Automotive AI is entering production after several years of experimentation. As this shift accelerates, OEM cybersecurity teams are expanding their mission beyond traditional vehicle security.

Programs that once focused on in-vehicle systems must now defend a connected mobility ecosystem where AI capabilities span onboard vehicle systems, cloud platforms, and connected applications. In fact, APIs have emerged as the nervous system that governs command & control and data.

In this webinar, Dan O’Reilly from Ford and Tomer Younger from Upstream examine how automotive cyber teams are evolving to address this shift.

The discussion will look ahead to risks introduced by AI-driven architectures. As large language models (LLMs) power vehicle features and operational workflows, and MCP orchestration enables dynamic tool usage across systems, cybersecurity teams must prepare for new attack surfaces and vulnerabilities. These include prompt injection, data leakage through model context, and resource exhaustion attacks targeting LLM infrastructure.

Key topics and takeaways:

  • How SOC operations are evolving to defend a connected mobility ecosystem shaped by AI agents
  • Why APIs have become the operational control plane for connected vehicle platforms and what effective security operations look like at fleet scale
  • How cybersecurity teams should prepare for emerging risks introduced by AI-driven systems, including LLMs and MCP

Watch the Recording