Particle.news

OpenAI Launches Lockdown Mode and Elevated Risk Labels for ChatGPT Enterprise

Admins gain tighter controls that keep browsing to cached content, with consumer access planned later.

Overview

  • Lockdown Mode is an optional setting that deterministically disables higher-risk tools and limits external interactions to reduce prompt-injection and data-exfiltration threats.
  • In this mode, web browsing relies on cached pages so no live network requests leave OpenAI’s secure network.
  • The feature is available now for ChatGPT Enterprise, Edu, Healthcare, and Teachers, with workspace administrators enabling it in Workspace Settings and selecting allowed apps and actions.
  • New Elevated Risk labels appear across ChatGPT, ChatGPT Atlas, and Codex to explain when features introduce extra exposure, such as granting network access in Codex.
  • OpenAI says these controls target users with heightened security needs and that labels may be removed as safeguards improve, with consumer availability planned in a later phase.