• Introducing Lockdown Mode and Elevated Risk labels in ChatGPT to help organizations defend against prompt injection and AI-driven data exfiltration.

Article Summaries:

  • OpenAI has rolled out two new security features for ChatGPT-Lockdown Mode and Elevated Risk labels-to help organizations guard against prompt‑injection attacks and AI‑driven data exfiltration. Lockdown Mode limits the model’s ability to access external data and perform certain actions, effectively tightening the sandbox around user interactions. Elevated Risk labels flag content that could trigger higher‑risk behaviors, allowing administrators to review or block potentially dangerous outputs. These updates are part of OpenAI’s broader effort to strengthen enterprise security and give customers more granular control over how the model handles sensitive information.

Sources: