OpenAI has published details on its forthcoming ChatGPT agent System Card, set for release on July 17, 2025. This new agentic model, part of the OpenAI o3 family, integrates advanced capabilities. It combines Deep Research's multi-step research and reporting with Operator's remote visual browser task execution.
The ChatGPT agent also features a terminal tool for code execution and data analysis. Furthermore, it generates slides and spreadsheets. It accesses external data sources like Google Drive via first-party Connectors. OpenAI prioritizes safety, building on Operator's robust controls. They added safeguards for broader user reach and terminal access risks.
Understanding ChatGPT Agent's Advanced Capabilities
OpenAI classifies this launch as "High capability" within the Biological and Chemical domain under its Preparedness Framework. Consequently, the company activates associated safeguards. While no definitive evidence suggests the model could help novices create severe biological harm, OpenAI adopts a precautionary approach. This proactive stance underscores their commitment to responsible AI deployment.
