Securing Autonomous AI Agents in Real Time with Norton 360

With autonomous AI agents gaining more power by the day, it is more important than ever to prioritise security.

Cybersecurity leader Norton has moved to address this growing security gap by introducing the beta version of Norton AI Agent Protection within Norton 360. 

Norton’s AI Agent Protection aims to help users interact with autonomous AI tools more safely as they begin handling sensitive tasks on their behalf.

AI agents are increasingly being used to manage work, automate decisions and access personal data across devices and accounts. 

While this brings clear productivity gains, it also introduces new risks that traditional cybersecurity tools were never designed to handle.

Unlike older forms of malware that simply corrupt files or systems, compromised AI agents can take action independently, which raises the stakes significantly.

“People are giving AI agents significant access to their machines, accounts and personal information because that's what makes them powerful," says Travis Witteveen, Head of Products and Portfolios at Gen, the parent company of Norton. 

“But until now, there's been no way to verify what those agents are about to do and the potential harm they could cause with one bad decision or click. 

“Norton AI Agent Protection addresses that missing trust layer, so people can use these tools with confidence, not guesswork.”

Real-time control for safer AI

At the centre of Norton’s new approach is Safe AI Execution, a system designed to monitor and assess every action an AI agent attempts in real time. 

Rather than allowing unrestricted autonomy, the feature introduces a checkpoint between intention and execution.

Safe actions are allowed to continue without disruption, while confirmed threats are automatically blocked before any damage can occur. 

When behaviour appears unusual or risky, the system pauses activity and prompts the user to review what the AI is attempting to do.

This model is designed to give users confidence in using agentic AI systems without constantly second guessing their actions. 

It also reflects a shift in cybersecurity thinking, where oversight moves closer to the moment decisions are made rather than reacting after harm has already occurred.

Rising threats 

The introduction of Norton AI Agent Protection comes at a time when AI ecosystems are expanding rapidly and so are the associated risks. 

According to Gen’s Threat Labs, there are already hundreds of malicious skills circulating within public agent registries. These can be used to manipulate behaviour, extract credentials or trigger harmful commands.

Even legitimate AI agents can create vulnerabilities if they are granted wide access to systems and data. 

Once compromised, they can operate quickly and independently, often without visible signs of malicious activity until it is too late for users to respond.

This evolving threat landscape is pushing security providers to rethink how protection is delivered.

Building the Agent Trust Layer

Norton AI Agent Protection has been developed by Gen Threat Labs and the Gen AI Foundry team, which is focused on scaling next generation AI driven security products. 

The launch marks a key milestone in the company’s wider ambition to establish what it calls the Agent Trust Layer, a framework designed to secure AI systems from decision making through to execution.

By introducing real time enforcement at the point of action, Norton aims to ensure users retain meaningful control even as AI systems become more autonomous.

 The goal is to make advanced AI tools usable without requiring technical expertise or constant oversight.

The feature is currently available in beta for Norton 360 customers on Windows, with Mac support expected to follow.

It is designed to work with popular AI development and automation tools including Claude Code, Cursor and OpenClaw, signalling Norton’s intent to support the environments where AI agents are increasingly being built and deployed.