Autonomous Vehicles leverage the abilities of AI and have converted cars into moving computers with sensors, cameras, and microphones operated by sophisticated software with heavy reliance on internet communication to acquire and process information such as speed, location, and required human-to-machine communications. It implies the possible data accessibility by skilled attackers for Malicious purposes, just like authorized individuals (manufacturers) access this information to boost safety and improve quality and performance.
Autonomous Vehicles Vulnerabilities And Threats
Because Autonomous vehicles are highly connected to the internet, it provides some opportunities for threat actors to leverage the need for integrating hardware and software interaction between systems to infiltrate and manipulate vehicle’s functionality. threat actors could manipulate systems such as internet connectivity, connectivity with other vehicles, navigation, security and other vital infrastructures.
An attack on the navigation system could help the threat actors manipulate the vehicle to misinterpret signs, speed without control, travel in the wrong direction and, in the worst cases, drive passengers to their deaths. It implies the Internet of Vehicles replicates security issues associated with the Internet of Things.
The world is a global village, and it operates a global market where products cross from one border to another. Already, we have some known autonomous car manufacturers with profound reputations when it comes to providing top-quality cars. Despite their reputation, we can’t deny the possibility of a sponsored attack through them. There could be a situation where a manufacturer leverages AI to develop an inbuilt spy feature that aids access and unauthorized data collection and transfer to an enemy or terrorist groups.
An example of such a spy attack is evident in a survey by CSIS, revealing 152 publicly reported instances of Chinese espionage directed at the United States since 2000. It’s difficult even for cybersecurity experts to reduce possible risks and damages associated with the importation and usage of such autonomous vehicles. Taking autonomous vehicle supplies from untrusted sources could also lead to a ransomware attack.
Despite promoting safe driving through automated and interconnected systems with a better understanding of driving skills, rules and signs than humans, autonomous vehicles can still be a significant security threat when threat actors succeed in taking over the systems using malware.
According to the Wall Street Journal, a car knows more about an individual than imagined. A smart car can spy, record or save information and the storage of the information and people/authorities who also have access to it should bring worries. According to the Chinese 2017 National Intelligence Law, the Chinese government is permitted to request information from Chinese companies and citizens, which implies the government could access the location and conversations of users of autonomous Chinese-made cars. Another concern regarding unauthorized data collection is the possibility of threat actors directing a copy of all data to an accessible location without hindering its transfer by the manufacturer to the intended database.
Autonomous Vehicles Security Measures
The security of every level of an AI system development for an autonomous vehicle is essential for several reasons, which include securing lives and protecting data. Factors such as encryption and access controls are necessary at all times. Also, it’s vital to understand that there is no such thing as “everlasting security,” The AI must regularly get updates to patch known vulnerabilities and prevent future ones.
Autonomous vehicles should have stringent network security, including firewalls, intrusion detection and prevention systems. Since there’ll be external system communication with the vehicle, secure communication protocols using data encryption methods are necessary.
“High Availability” is one of the vital factors to keep an AI system functioning. It implies encouraging Redundancy to prevent a single point of failure and promote the detection of sensor spoofing or failures. However, data availability without adequate data validation could be a recipe for disaster. AI systems’ design should cross-check and validate data from all sources at all times.
When things go wrong with an autonomous vehicle, except it’s only an attempt to gather information from its database, one can detect it through a new behaviour in the system functionality. Implementing real-time behaviour monitoring can help identify anomalies in a vehicle’s operation and trigger alerts in case of unauthorized access or tampering. IDS and IPS will be very helpful in doing this.
Governments and regulatory bodies should establish standards and regulations for developing, deploying, and operating autonomous vehicles to ensure safety and security. It will prevent the release of substandard autonomous vehicles to the market for people’s use.
We live in a world where people use technologies without reading privacy policies, terms and conditions, etc. Nevertheless, it should be mandatory for passengers and operators of autonomous vehicles to pass through some basic training on risks and safety measures. It will help them understand safe use guidelines and security best practices.
Conclusion
The security implications of AI in autonomous vehicles are complex and evolving. As technology continues to advance, so do the strategies of potential attackers. Ensuring the security of self-driving cars is an ongoing process that requires collaboration between automakers, software developers, cybersecurity experts, and policymakers. The promise of safer, more efficient transportation through autonomous vehicles is within reach, but only if we address the security implications with vigilance and commitment. These resources by CISA are a useful tool in gaining more knowledge about securing autonomous vehicles.
