PhantomLiDAR Attack Exposes Vulnerabilities in Autonomous Vehicle Sensors

At the NDSS 2025 symposium, researchers from Zhejiang University introduced PhantomLiDAR, a novel attack method that exploits vulnerabilities in LiDAR systems essential for autonomous driving. This attack involves injecting intentional electromagnetic interference (IEMI) to disrupt LiDAR outputs, showcasing a new dimension of threat beyond traditional laser signal attacks.

The study identified several attack surfaces within LiDAR systems, including the laser receiving circuit and beam-steering modules. The researchers proposed an alias attack that can manipulate LiDAR outputs in various ways, such as Points Interference, Points Injection, Points Removal, and even complete Power-Off of the LiDAR system.

Through both simulated and real-world experiments, the researchers demonstrated the effectiveness of these attacks on five COTS LiDAR systems. They also conducted feasibility tests in moving scenarios, indicating that these vulnerabilities could be exploited in practical applications.

This research underscores the importance of understanding the risks associated with electromagnetic interference in autonomous vehicle technology. As LiDAR becomes increasingly integral to vehicle navigation and safety, the implications of such attacks could be severe, potentially leading to accidents or system failures.

Organizations developing autonomous vehicles should take these findings seriously and consider implementing defense measures at both the sensor and vehicle system levels. This includes enhancing electromagnetic compatibility testing and developing robust countermeasures against IEMI attacks.

As the technology evolves, staying informed about emerging threats like PhantomLiDAR is crucial for manufacturers and users alike. Monitoring advancements in LiDAR technology and associated security measures will be essential to mitigate these risks effectively.