ChatGPT Just Got a Lockdown Mode — Here's What It Does and Who Needs It

OpenAI quietly dropped a significant security update over the weekend. As of February 16, 2026, ChatGPT now has a Lockdown Mode — an optional setting designed to protect high-risk users from prompt injection attacks.

Let's break down what it does, who it's for, and why it matters.

---

What Is ChatGPT Lockdown Mode?

Lockdown Mode is a new security setting that tightly restricts how ChatGPT interacts with external systems. The goal? Prevent attackers from tricking ChatGPT into leaking your data through prompt injection.

Here's what it does:

• Web browsing is limited to cached content — no live network requests leave OpenAI's servers, so malicious websites can't exfiltrate your data
• Certain tools and capabilities are disabled entirely when OpenAI can't guarantee data safety
• Deterministic protections — these aren't AI-based filters that might fail, they're hard restrictions baked into the system

Who Is It For?

OpenAI is clear: Lockdown Mode is not for most users. It's designed for:

• Executives at prominent organizations
• Security teams handling sensitive information
• Anyone at higher risk of targeted cyberattacks
• Users working with confidential data in connected apps

Think of it like Apple's Lockdown Mode for iPhones — extreme protection for people who genuinely need it.

What Are "Elevated Risk" Labels?

Alongside Lockdown Mode, OpenAI also introduced Elevated Risk labels across ChatGPT, ChatGPT Atlas, and Codex. These labels flag capabilities that could introduce additional security risk, giving users and IT admins clearer visibility into what's happening.

Why This Matters in 2026

Prompt injection has become one of the biggest security headaches in AI. As ChatGPT connects to more external tools — browsing, code execution, third-party plugins — the attack surface keeps growing.

A malicious website or document could embed hidden instructions that trick ChatGPT into:

• Sending your conversation data to an attacker
• Executing unintended actions in connected apps
• Revealing sensitive context from your chat history

Lockdown Mode is OpenAI's first deterministic defense against this. Not a model-level fix that might be bypassed — a hard system-level restriction.

How to Enable It

Lockdown Mode is available for:

• ChatGPT Enterprise
• ChatGPT Edu
• ChatGPT Team (rolling out)

Admins can enable it through workspace settings. Free and Plus users don't have access yet.

The Takeaway

For most people, ChatGPT's existing security is fine. But if you're handling sensitive business data or you're a potential target for cyberattacks, Lockdown Mode is worth enabling immediately.

It's a smart move by OpenAI — and a sign that AI security is finally getting the serious attention it deserves.

---

Want to get more out of ChatGPT? Check out our best AI prompts collection to boost your productivity, or compare the top AI chatbots of 2026 to find the right one for you.