Norton Adds VPN and Guardrails for AI Agents — Here’s What It Means for Your Privacy

If you’ve ever asked ChatGPT for help drafting an email, used Microsoft Copilot to summarize a meeting, or let an AI assistant read your calendar to suggest a schedule, you’ve handed over information that may not stay as private as you think. Most cloud-based AI agents send your prompts—and sometimes your personal data—to remote servers for processing. That traffic is typically encrypted in transit, but the platform provider still sees your input, and third parties (like your internet service provider) can see that you’re communicating with an AI service.

On April 30, 2026, Gen Digital (the parent company of Norton, LifeLock, and Avira) announced two new features aimed at mitigating those risks: a VPN specifically designed to encrypt traffic to and from AI agents, and on-device guardrails that block sensitive information from being sent in the first place.

What Happened

Gen’s announcement, carried by PR Newswire and reported by Stock Titan, introduces:

  • A VPN for AI agents – This is not a general-purpose VPN. It creates an encrypted tunnel specifically for communications with AI services. According to the press release, this prevents anyone on your local network (including your ISP) from seeing which AI tools you use or intercepting your prompts.
  • Guardrails in Norton – A local, on-device AI feature that scans your prompts before they leave your computer or phone. It can detect and block common sensitive data such as Social Security numbers, credit card numbers, and passwords, and it alerts you before the information is sent to an AI agent.

The features are rolling out to Norton 360 subscribers first. Gen has not yet announced pricing for a standalone product, but present subscribers should be able to activate these options within their existing dashboard.

Why It Matters

For everyday users, the key risk with AI agents is not just that a company might misuse your data—it’s that you might accidentally paste something you didn’t intend to share. A survey by the Pew Research Center found that a sizable portion of users have typed sensitive information into chatbots without realizing it could be stored or reviewed.

The guardrails address that directly: they work on your device, so even if you’re about to paste a bank account number into a prompt, the software can block it and ask for confirmation. That’s a meaningful improvement over relying on the AI provider’s own privacy policies.

The VPN for AI agents is a more subtle benefit. A regular VPN already encrypts all your traffic, but Gen’s approach is narrower: it only routes AI service traffic through a dedicated encrypted tunnel. That means less latency for other activities, and it may be easier to configure than a full VPN (you don’t have to toggle it on globally). If you work from a coffee shop or use public Wi-Fi frequently, this extra layer ensures that any prompt you type into an AI assistant isn’t visible to others on your network.

What You Can Do

If you already have a Norton 360 subscription, here’s how to take advantage of these features:

  1. Check for updates – Norton automatically updates, but you may need to restart the app or manually check for the latest version (look for a version released after April 30, 2026).
  2. Find the AI security section – In the Norton dashboard, look for a new tab labeled “AI Security” or “Agent Security.” This is where the VPN toggle and guardrail settings live.
  3. Enable the VPN for AI – Turn on the VPN switch. By default it covers most major AI agents (ChatGPT, Copilot, Gemini, Claude, Perplexity). Norton’s support page likely lists all supported services.
  4. Set guardrail sensitivity – You can choose which types of data to block. Most users should leave “High” sensitivity enabled, which blocks common account numbers and personal identifiers. A lower setting may only flag certain categories.

If you don’t have Norton, the company has indicated a standalone product may be available later this year. In the meantime, you can achieve partial protection by using a general VPN while interacting with AI agents, but you’ll miss the guardrails—and those are arguably the more valuable feature for avoiding accidental leaks.

Sources

  • Gen Digital press release, “Gen Accelerates Agentic Security and Privacy for the AI Era,” PR Newswire, April 30, 2026.
  • Stock Titan, “Gen rolls out a VPN for AI agents and adds guardrails in Norton,” April 30, 2026.
  • Pew Research Center, “The Internet and the Pandemic,” 2023 (related data on sharing sensitive info online).