Academic security researchers Sandeep Gupta and Roberto Passerone published their research paper a couple of months ago, summarizing attacks, vulnerabilities, and risks in existing vision systems for modern cars.
Vision systems contain cameras, LiDAR, RADAR, ultrasonic sensors, and AI models, and the authors review all the previously mentioned attack vectors that affect these systems in real cars. So you have a summary and all the links up to September 2026 in one place.
Great summary with many interesting examples (like how to turn a 30 km/h sign into an 80 km/h sign with a flashlight, or how to hide a person with a T-shirt print). If you work with autonomous systems, this is definitely a good read for the weekend.
Enjoy, and please share with friends and colleagues in the automotive industry. :)
More details:
Robust Vision Systems for Connected and Autonomous Vehicles: Security Challenges and Attack Vectors [PDF]: https://arxiv.org/abs/2602.09740