On the security and privacy of LLMs in Mobility
Survey finds LLM mobility research largely neglects security, privacy, and EU AI Act compliance despite transportation AI's high-risk classification.
The paper surveys LLM applications in the mobility sector (a roughly $2.9 trillion annual market affecting over 1.5 billion vehicles) and derives nine technical requirement classes from the EU AI Act, which classifies transportation AI as high risk. Reviewed research focuses mainly on GPT and Llama models (over 50% of works) and traffic applications, while largely neglecting security, privacy, and reliability. Of 35 reviewed works, only one includes a partial vulnerability assessment and one a partial risk management system, revealing a large gap between optimization performance and regulatory adherence.
- Nine technical classes derived from EU AI Act high-risk requirements
- GPT and Llama dominate over 50% of reviewed mobility LLM research
- Only 1 of 35 works includes a partial vulnerability assessment
- Authors urge security-by-design for safety-critical intelligent transportation systems
From the original article
1 min read at arxiv.orgThe mobility sector is undergoing a paradigm shift driven by advances in Generative Artificial Intelligence. With a global market valued at approximately 2. 9 trillion dollars annually, considering only cars, the integration of these technologies has the potential to impact more than 1.
By Mauro Conti · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.