June 18, 2024 Federal officials, AI model operators, and cybersecurity companies conducted the first joint simulation of a cyberattack involving a critical AI system. The exercise, hosted by the Cybersecurity and Infrastructure Security Agency’s (CISA) Joint Cyber Defense Collaborative (JCDC), aimed to address the unique challenges posed by AI-related cyber threats.

Responding to cyberattacks on AI-enabled systems requires different strategies compared to typical hacks. As AI tools become more prevalent, hackers could exploit them to enhance and scale their attacks. This simulation is part of broader efforts by Washington and Silicon Valley to preemptively tackle these evolving threats.

The tabletop exercise took place at Microsoft’s offices in Reston, Virginia. While specific details of the simulated incident were not disclosed, the focus was on current threats and how information can be shared between the government and private sector.

More than 50 AI experts participated, including representatives from Amazon Web Services, Microsoft, Nvidia, OpenAI, and Palantir.

Kyle Wilhoit, director of threat research at Palo Alto Networks’ Unit 42, highlighted that the exercise provided an opportunity to discuss current threats and hypothesize future attack vectors leveraging AI. The exercise also helped CISA identify the appropriate contacts in the private sector for AI-related incidents.

The insights gained from the exercise will inform CISA’s forthcoming AI security incident playbook, expected to be published by the end of the year. The JCDC plans to host another AI tabletop exercise before the playbook’s release.

The first joint AI attack simulation marks a significant step in understanding and addressing the unique cyber threats posed by AI systems. The collaboration between federal officials, AI model operators, and cybersecurity companies underscores the importance of proactive measures and information sharing to safeguard AI technologies.