IRONSCALES: 77% of firms saw AI agents act on malicious content

An IRONSCALES report based on a Censuswide survey of 501 U.S. IT and security professionals found that 77% saw an AI agent act on malicious, misleading or unauthorized content in the past year, with 57% confirming at least one incident. Agents had access to ticketing systems at 67% of firms, email at 49% and finance systems at 30%.

• 77% reported AI agents acting on malicious content in the past year
• 57% confirmed at least one AI-agent-related incident
• 67% give agents access to ticketing, 30% to finance systems
• Only 6% say security leads approval of AI tools