Cyber and AI: Balancing Innovation and Responsibility

Cyberattacks are increasing, putting the nation at risk and pushing the government to find new, advanced ways to defend systems and networks.

In FY 2023, federal agencies reported 32,211 attacks on critical infrastructure and devices, up 9.9% from the previous year. Most attacks follow standard tactics, but threat actors increasingly use artificial intelligence and generative AI to make incidents harder to track, automating and customizing attacks for a more targeted approach. Generative AI models, for example, allow adversaries to create sophisticated phishing attacks using social media accounts. The quick pace of AI development also makes securing networks more challenging. AI lets adversaries hit more people faster and more pervasively; this doesn’t make the adversaries smarter — they just have better tools. To counter this, federal agencies can use adversarial AI to detect and defend against these threats but first, they must work with industry to define what secure AI models look like.