OpenAI expands Lockdown Mode to all ChatGPT users

OpenAI has introduced a new security setting called Lockdown Mode for ChatGPT. The feature limits the model’s ability to access live web browsing, external connectors, file downloads and other network‑enabled tools, aiming to reduce the risk of prompt‑injection attacks that could exfiltrate sensitive data. OpenAI says the mode "reduces the number of doors ChatGPT can use to talk to the outside internet" and blocks all outbound network requests while a conversation is active.

Previously available only to paying enterprise customers, Lockdown Mode is now offered to every logged‑in user, including free‑account holders and Plus subscribers. The rollout also adds an improved session‑overview panel that lets users see and remotely terminate active logins. OpenAI notes that the setting does not filter malicious instructions but merely prevents data from leaving the model.

The announcement was covered by technology outlets in South Africa and Pakistan, highlighting the relevance of the new tool for high‑risk users such as executives, lawyers, healthcare workers and journalists.