Unchecked AI Agents: A Growing Security Threat

Many businesses are struggling to effectively monitor and manage the AI agents already active in their operations. This concern comes from Samantha Rule, a CISO at Ninety One. Here's the thing: these are not just chatbots. We're talking about AI agents with autonomy, capable of acting on information given to them. What's interesting is that the average enterprise has about 37 deployed agents, but only 24% have full visibility of agent-to-agent traffic. This lack of oversight means 88% of organizations reported an AI agent security incident in the last year. The bottom line is a significant security gap exists, with 41% to 44% of organizations lacking human control over high-risk agent decisions. To address this, organizations should treat agents like people, apply the same identity and accountability rules, and provide clear proof for every privileged action. It's also crucial to record all actions and install safeguards against irreversible decisions. This matters because unchecked AI agents pose a real and growing security risk to businesses.