OpenAI today announced the launch of two new protective measures aimed at helping users and organizations mitigate the risks of prompt injection attacks, while providing clearer visibility and stronger control capabilities.

These two new features include:

- **Lockdown Mode in ChatGPT**: This is an advanced optional security setting designed for high-risk users.

- **Elevated Risk Labels**: For certain features in ChatGPT, ChatGPT Atlas, and Codex that may introduce additional risks, the system will mark them with "elevated risk" labels.

Lockdown Mode allows users to limit ChatGPT's direct internet exposure, thereby reducing potential security threats. Meanwhile, the elevated risk labels clearly identify certain functions that may pose risks in the user interface or tools, such as external tool access.

Some netizens commented that these measures are particularly beneficial for Codex users, providing more practical protection. However, some users have expressed dissatisfaction with certain OpenAI policies and are even considering canceling their subscriptions.

Despite the controversy, the launch of these two new features demonstrates OpenAI's ongoing efforts in security. For developers and enterprise users, these tools may help better manage the risks associated with AI model usage.

For more detailed information, please refer to the [OpenAI official announcement](https://openai.com/index/introducing-lockdown-mode-and-elevated-risk-labels-in-chatgpt/).