S6E1: Resilient Cyber w/ Rob van der Veer

In this season premiere of Resilient Cyber, Chris Hughes sits down with Rob van der Veer, a seasoned AI and software security leader, to explore the evolving AI security landscape. With over 30 years of experience in AI, Rob shares his insights on the risks and opportunities presented by AI systems, focusing on LLMs (Large Language Models), AI supply chain concerns, and the increasing need for regulation.

Key Highlights:

• Rob's background in AI development, consultancy, and contributions to major projects like the European AI Act and ISO standards.

• Discussion on AI security risks, including purpose alignment, training data manipulation, prompt injection, and confidentiality risks in LLMs.

• The AI supply chain complexity, including the use of pre-trained models and external data sets.

• Regulatory approaches to AI, comparing the EU AI Act and the US voluntary frameworks.

• Insights into open source AI security initiatives like OWASP AI Exchange and OpenCR, which provide guidance on AI threats, controls, and cross-mapping compliance frameworks.

• The growing importance of cyber resilience and how AI systems need to incorporate zero trust principles to manage risks effectively.