Press release · November 17, 2026 · prnewswire.co.uk
Cato Networks introduces AgentSense to assess AI agents
- About Cato Networks
- Also names Influence
- It
- Read the original
Summary
Cato Networks has introduced Cato AgentSense, a patented capability from Cato AI Labs that assesses AI agents in real time. It is designed to protect the integrity of agent decisions, actions, and outputs, keeping enterprises in control as agents gain autonomy.
AI agents are scaling faster than the guardrails designed to secure them. According to Gartner, 71% of organizations expect to deploy agentic AI within the next 12 months, while only 21% currently have agentic AI governance capabilities in place. The release also cites forecasts that ungoverned AI agent abuse will drive 25 percent of enterprise breaches by 2028.
Agents cannot reliably separate trusted instructions from untrusted data encountered during work, and attackers can exploit that gap by hiding instructions in emails, webpages, documents, tickets, or tool responses. AgentSense Untrusted Influence Detection traces how instructions and data shape agent behavior at runtime and blocks an unsafe action before execution when untrusted content steers the agent toward it.
AgentSense Drift Detection will identify when agent actions depart from the original task or intent, evaluating whether each step still serves the intended objective as a workflow progresses. It complements Untrusted Influence Detection by assessing whether actions remain aligned with an agent's purpose. AgentSense is integrated within the Cato platform and included at no additional charge with Cato AI Usage licenses.
Cato also introduced the AgentSense Challenge, inviting the security community to attack an AgentSense-protected agent workflow with planted instructions. The first three eligible participants to complete it will each receive $10,000 USD. “As enterprises entrust AI agents with decisions that shape revenue, customer relationships, and business continuity, control over agent behavior has become as essential as financial controls,” said Shlomo Kramer, CEO and co-founder of Cato Networks.
“AgentSense helps preserve that trust by tracing how untrusted content influences an agent's decisions and sensitive actions, helping enterprises adopt AI agents without giving attackers control of the workflow,” said Itay Ravia, head of Cato AI Labs. Cato plans to use participant findings and feedback to evaluate and strengthen the capability.
Our summary. The original keeps its own byline.