# OpenAI Releases Safety Documentation for Operator AI Agent

OpenAI has published a System Card detailing the safety measures built into Operator, its new AI agent capable of performing tasks on the web.

The document outlines a comprehensive security approach addressing key concerns around AI agents that can interact with websites and applications. According to the announcement, OpenAI has implemented multiple layers of protection against prompt injection attacks and jailbreak attempts—techniques that could trick the AI into performing unintended actions.

The System Card also covers privacy and security safeguards designed to protect user data as the AI navigates websites and handles sensitive information. This is particularly critical since Operator can perform actions like filling out forms or making purchases on behalf of users.

OpenAI emphasized its external red teaming efforts, where security researchers attempt to find vulnerabilities before public release. The company included results from safety evaluations and acknowledged that refining these protections is an ongoing process.