New ChatGPT Lockdown Mode Limits Tools That Could Enable Data Exfiltration

OpenAI's Lockdown Mode is an opt-in security feature for ChatGPT that limits or disables tool capabilities (outbound web requests, third-party tool calls, networked code execution, external image rendering, markdown link rendering) to reduce the risk of data exfiltration from prompt injection attacks; the report explains target users, how to enable it, its limitations, and situates the feature within a maturing AI security ecosystem.