Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Cato Networks unveils AgentSense to stop harmful AI agent behaviour in real time

Cato Networks has unveiled AgentSense, a capability it says stops harmful AI agent behaviour in real time, presented as a patented breakthrough. The product puts protection at the runtime layer, judging what an agent does rather than only filtering what goes in.

Published

Cato Networks has unveiled a capability called AgentSense, which it says stops harmful AI agent behaviour in real time and describes as a patented breakthrough.

The announcement reached readers through VMblog's report, and the emphasis sits on two claims, real-time blocking and a patent, rather than on deployment figures, named customers or implementation detail.

The positioning is worth noting because agentic products are being handed more and more authority: calling tools, reaching into internal systems and carrying out processes on a person's behalf. Protection that only filters prompts at the input struggles to cover the steps an agent decides on during a run, and auditing after the fact usually comes too late to limit the damage.

AgentSense targets that runtime window, judging behaviour as it happens rather than blocking at the door or reconstructing events afterwards. Putting the patent language into the launch framing also suggests the vendor wants the technique treated as a differentiator rather than a routine security feature update.

Public detail remains thin. How harm is defined, how false positives and misses are balanced, and whether existing network and security policies need reworking are all unanswered. For enterprise buyers, independent testing and performance in live environments will carry more weight than the patent claim.

The backdrop is that more companies are wiring agents into ticketing systems, email, databases and code repositories. Once an agent is steered by a prompt injection or a bad instruction, the damage travels along whatever permissions it was granted, which turns runtime behaviour control from a nice-to-have into a precondition for deployment.

Three things to watch: general availability and product form, whether it ships as a standalone offering or a module inside an existing platform; any third-party testing or customer references; and how it integrates with the agent frameworks teams already use.

Why it matters

As agents gain tool and system access, security shifts from input filtering to runtime behaviour control. If AgentSense delivers real-time blocking in practice, it addresses a central concern in enterprise agent deployment.

Cato NetworksAI SecurityAI Agent
Back to realtime news

Nearby Updates

All

11/18, 02:10

Sqwish Labs offers decision-model customization trained on customer data

Sqwish Labs is offering customization for decision models, letting customers train the system on their own data to extend what it can do, TechCrunch reported on November 17. The launch is framed against a backdrop of rapidly rising decision-model adoption.

11/18, 02:04

Pedestal AI raises $4.7M to put agents to work inside supply chains

Pedestal AI has raised $4.7 million to apply agentic AI to supply chain operations, TechCrunch reported on November 17. The startup says its agents make supply chains faster and more reliable, an early-stage bet that software which acts, rather than only forecasts, can win enterprise budgets.

11/18, 01:57

Anthropic opens linked billing beta so Claude API apps can charge their own customers

Anthropic launched linked billing in beta on November 17, letting developers charge Claude API usage in their applications to a customer's own Claude organization. Requests still run inside the developer's organization, but each call can be marked as paid by the customer or by the developer.

11/18, 01:45

OpenAI launches GPT-6.2 Astra and Sol for professional work, coding and computer use

OpenAI introduced GPT-6.2 Astra and Sol on November 17, describing better professional outputs, coding, computer use and efficiency at lower cost across ChatGPT Work, Codex and the API. Both models arrive in one announcement without a stated division of labour, so the practical differences will only surface in documentation and pricing.