National · Cybersecurity

CISA and partners issue guide for careful adoption of agentic AI

The joint guidance advises organizations to begin with lower-risk uses and limit broad access as they adopt AI services that can take actions.

A developer working with AI-assisted code on a laptop.

CISA, Australia's ASD ACSC, and international partners released Careful Adoption of Agentic AI Services on May 1. The guide highlights risks that can arise when AI systems act through connected tools and recommends staged, security-led adoption.