I’m working through a threat model for a fleet of vehicles that use dash cams, and I keep running into the data-handling side rather than the hardware. The cams record continuously, store footage on an SD card, and some models upload clips to a cloud account. I can see the usual risks around someone pulling the card, weak account passwords, and unencrypted uploads, but I’m not confident I’ve covered what happens on the vendor side or how long clips are retained. I also don’t know whether motion-triggered recording changes the trust boundary in any meaningful way. If you’ve modeled similar camera or IoT setups, I’d like to know which threats you prioritized and whether you treated the cloud service as a separate trust zone. What has actually been useful in practice?
Found Best Dash Cam of 2026 while reading up on retention and vendor handling, and it reminded me I still haven’t settled whether the cloud service should be its own trust zone. I’m leaning that way, but I’m unsure how much motion-triggered recording actually shifts the boundary.
Good for you: the 2025 TMC Hackathon was about self-driving cars. I bet they have covered similar topics. Maybe there’s something inspiring for you in there.
I would add: That sounds like a tough problem. And security with all of it’s information disclosure, availabilty (what if sensors break down?) is probably not your biggest concerns. You should add Usability Threat Modeling to the mix. What are you trying to accomplish? What could make that fail? (Oops, I didn’t see that pedastrian / …) How can you make it succeed?