- OpenAI introduced Lockdown Mode to reduce prompt injection risks when ChatGPT handles web pages, files, or external content.
- The feature restricts high-risk tools like live browsing, web image retrieval, deep research, and agent mode to limit data leakage.
- It targets sensitive-data users, including organizations, executives, security teams, and eligible ChatGPT accounts, but does not fully eliminate risk.
OpenAI has introduced Lockdown Mode, a new security feature designed to reduce the risk of prompt injection attacks in ChatGPT and supported…
