Revolutional cyber lead urges accountability for federal AI tools

Pete Harris, head of cyber operations at Revolutional, said AI systems used in cybersecurity need clear policies, regulation and accountability as federal agencies expand their use. Speaking at Cyber Leaders Exchange 2026, he warned that algorithms can make consequential mistakes if agencies cannot trace how decisions were made or assign responsibility for outcomes.

Federal employees are mainly using AI to speed routine work such as data analysis and report generation, while security teams are testing bots for narrow tasks like threat intelligence collection, activity log review and security operations center analysis. Harris said analysts may eventually act as “AI commanders,” overseeing fleets of tools, including scenarios where one analyst manages 8,000 AI bots and audits selected agents through documented trails.

Unapproved tools remain a major concern. Harris pointed to workplace apps adopted 10 years ago that may now include AI chatbots with access to organizational data, along with employee use of Copilot, Claude, ChatGPT and other tools on personal devices. He said agencies should press vendors on failures, hallucinations, audit trails and damage control, especially when products claim autonomous security capabilities. If an AI system reacts at 2 a.m. and gets it wrong, agencies need to understand the consequences before deployment.