- Chief information officers warned of the potential power of artificial intelligence agents to conduct cyberattacks
- The CIOs shared their thoughts at the Department of Defense Intelligence Information System Worldwide conference
- Defense Information Systems Agency CIO Roger Greenwell said that intelligence experts are working on patching capabilities for future AI-enabled disruptions
Chief information officers across the intelligence community say they are bracing for a future in which artificial intelligence agents — not human hackers — carry out the next generation of major cyberattacks, GovCIO Media & Research reported on Monday.
The CIOs expressed their concern during a panel at the Department of Defense Intelligence Information System Worldwide conference in Tampa, Florida, where officials described how AI is simultaneously accelerating adversary capabilities and forcing agencies to rethink how they defend their networks.
Where AI, cyber and intelligence collide, the 2026 Intel Summit puts the National Security Agency, National Reconnaissance Office, FBI and Space Force leaders in the room to unpack what’s next. Reserve your seat for September 24 and hear firsthand how the intelligence community plans to out-innovate the threats reshaping its mission.
What Did the DIA CIO Say About Autonomous AI Agents as Threats?
Defense Intelligence Agency Chief Information Officer Edacheril Mathew pointed to recent incidents at three major AI companies as a case study worth close attention. In those episodes, sandboxed AI systems undergoing cybersecurity testing broke out of their containment environments to independently seek additional information from each other, according to Mathew. He said the systems went further, independently exchanging information with other AI systems.
Mathew said the episodes reinforce his view that future large-scale cyberattacks may increasingly be executed by autonomous agents rather than humans.
The risk is heightened by how interconnected intelligence agency networks already are. Mathew noted that roughly 600 organizations connect to the Joint Worldwide Intelligence Communications System, and he argued that agencies deploying AI across such environments need clearly defined limits on what autonomous systems are permitted to do — without which, actions by one system could ripple across every connected network.
What Are the Actions Being Taken By Intelligence CIOs?
Roger Greenwell, director of enterprise integration and innovation and CIO of the Defense Information Systems Agency, said adversaries are already using widely available AI tools to find vulnerabilities in U.S. systems more quickly than before. That is pushing federal agencies to detect and patch weaknesses at a much faster pace.
The rising volume and speed of cyber threats adds pressure on IT teams and raises the risk of service disruptions during a crisis, Greenwell said, adding that national security missions cannot tolerate outages. He said the intelligence community is working with industry partners to build capabilities that can patch systems dynamically as new vulnerabilities surface.
National Geospatial-Intelligence Agency CIO Mark Chatelain said his agency has successfully woven AI into its core mission work, but applying the technology to cyber defense brings its own set of hurdles. That includes using AI to spot threats hidden in network telemetry and building systems capable of responding to malicious activity within microseconds.
Office of the Director of National Intelligence CIO Douglas Cossa cautioned that technology alone won’t resolve agencies’ cybersecurity challenges, noting that many IT problems trace back to organizational culture rather than technical shortcomings. Unauthorized or uncoordinated changes made by employees are one recurring example, he said, and strong change management practices can help agencies prevent such actions while accounting for the impact of modifications on interconnected system’ components.
