The U.S. Department of Homeland Security sees three areas of concern as artificial intelligence is used across critical infrastructure sectors: attacks using AI, attacks targeting AI systems and design, and implementation failures.