OpenAI has outlined the mechanisms it employs to protect user data when AI agents interact with external links. These built-in safeguards are designed to prevent potential risks such as URL-based data exfiltration and prompt injection attacks.
Source: OpenAI