Critical Infrastructure Cybersecurity Summit

What happens when an AI system supporting an essential service produces an unsafe recommendation, is manipulated or fails during a cyber incident? Traditional incident response plans usually say who handles a network intrusion. However, they often do not clarify who can pause an AI system, override its output, check its impact or decide when it can return to service.

In this session, Jennifer Raiford presents an executive approach to AI incident response for critical infrastructure. She explains how to set up a clear chain of command that includes security, operations, AI governance, legal, and executive teams. She also focuses on keeping people in charge of important decisions and rehearsing response before a failure affects the public.

Key takeaways:
- Recognize when an AI failure or misuse requires incident escalation.
- Assign authority to pause, override, investigate, and restore an AI system.
- Coordinate cyber response, operational continuity, and public communication.
- Exercise AI incident scenarios before they affect essential services.