Researcher, Agent Safety, Oversight and System Mitigations
About this role
Professionals who enjoy tackling ambiguous safety questions by turning them into concrete threat models and practical mitigations will find a strong fit here. The Agent Safety team builds mechanisms to ensure increasingly capable artificial intelligence agents act safely, exercise sound judgment, and remain aligned with user intent.
Team members focus on oversight and system-level mitigations, designing controls for agent actions such as agent-based review, sandboxing with process isolation, and permission boundaries. The work involves partnering with a Codex harness engineering team to productionize these controls and red-teaming end-to-end agentic systems to prevent data exfiltration and unsafe tool use.
What you will do
- Design, build, and evaluate system-level controls for agent actions such as agent-based review.
- Plan how controls fit into a broader system that includes sandboxing with process isolation and permission boundaries.
- Work closely with a Codex harness engineering team to productionize AI controls.
- Red-team end-to-end agentic systems to measure whether controls prevent data exfiltration, unsafe tool use, and other harmful outcomes.
- Improve the safety and productivity tradeoff by measuring and reducing missed harmful actions, unnecessary blocks, approval burden, and latency.
What you need
- Possess strong systems or security instincts to reason concretely about isolation boundaries, permissions, attack surfaces, and failure modes in complex systems.
- Translate ambiguous safety questions into concrete threat models, reproducible experiments, and practical mitigations.
- Revise approaches based on evidence gathered from deployment.
- Build robust experimental infrastructure and design evaluations that distinguish promising mitigations from brittle ones.
- Demonstrate deep interest in frontier artificial intelligence alignment, safety, and control.
Nice to have
- A background in AI control or security
Benefits
- Offers Equity
- Relocation assistance to new employees
Skills and tech stack
Questions candidates ask
What is the remote work policy for this position?
The role is based in San Francisco, California. The team uses a hybrid work model requiring three days in the office per week.
What are the compensation details for the role?
The salary ranges from $380K to $500K and includes offers of equity.
What relocation support is available?
OpenAI offers relocation assistance to new employees joining this team.