
Introducing Cato AgentSense: a patented breakthrough in agentic security from Cato AI Labs that stops harmful agent actions in real time, before they execute.

Networking + Security, built to work as One. Cato delivers the converged cloud for the AI era: visibility, context, and control in one architecture.

Introducing Cato AgentSense: a patented breakthrough in agentic security from Cato AI Labs that stops harmful agent actions in real time, before they execute.

Watch the reveal: bit.ly/4Awxvi4 Explore AgentSense: bit.ly/4fzqdlu

When AgentSense detects untrusted content steering an agent toward an unsafe action, it blocks that action before execution, without relying on known attack patterns. Put AI agents to work with confidence.

Valid permissions don’t stop an agent from following an attacker’s instructions hidden in a document, email, or tool response. Built by Cato AI Labs, AgentSense traces how instructions and data influence an agent’s behavior.

@CatoCTRL’s Guy Waizel and Etay Maor examine what was being requested, what was circulating and what remained unverified. Read the full research: catonetworks.com/blog/cato-c…

For defenders, those doubts raise questions about prerequisites, reliability and the evidence behind an exploit claim. Public analysis discussed in the report examined how PHP version, server settings and target knowledge could affect some circulating PoCs.

Attackers are searching for a PHP2Shell exploit that actually works. In an underground forum, participants requested exploits and questioned the code being shared. One assessment: "Didn't test but seems legit."

You get more confidence. More control. To put AI agents in charge. See how. Live. Nov. 17. Register for the reveal: bit.ly/4BSCE5F


In his latest @Forbes article, Cato’s Etay Maor examines how AI is changing the speed and scale of ransomware operations, separating documented uses from sellers’ unproven claims.

They’re using it to map networks, identify valuable data, and negotiate with victims. Running the operation still takes skill. AI reduces the manual work that limits how much one operator can handle.

AI is helping ransomware operators run more attacks at once and keep extortion conversations going around the clock.

Cato also automatically blocks confirmed campaign indicators, without requiring customers to add those indicators manually. Read the Cato CTRL research for the delivery chain, warning signs, and Cato’s protection: bit.ly/4efYsOx

The lesson extends beyond this Israeli campaign: legitimate remote-support software can be misused when employees are tricked into running and installing it. Cato’s intrusion prevention system blocked a web request to infrastructure used by this campaign.

If someone runs the disguised program and it starts successfully, an attacker could gain remote access to the computer. The investigation found no evidence that this campaign exploited a SimpleHelp vulnerability or compromised SimpleHelp’s own infrastructure.

Recipients are directed to a lookalike website, asked to complete a verification step, and offered a ZIP download. Inside is a program disguised as a tax document. It uses SimpleHelp, legitimate remote-support software, configured to connect to attacker-controlled infrastructure