Researcher, AI Agent Safety & Controls

CHEManager International; San Francisco (CA); Hybrid; USD 180,000 - 260,000; Full time

Benefits offered by this job

Relocation assistance

Hybrid work model

Job summary

OpenAI seeks a safety-focused researcher/engineer in San Francisco to design, implement, and test system-level controls for increasingly capable AI agents in real-world environments. The role emphasizes building practical mitigations, rigorous threat models, and evaluation infrastructure, with a strong focus on safety and alignment.

The role supports a hybrid work model, collaborating with Codex teams and other security specialists to advance responsible deployment of AI systems.

Qualifications

• Experience designing system-level controls for AI or software systems.
• Ability to reason about isolation boundaries, permissions and attack surfaces.
• Experience building evaluative infrastructure and experiments to test mitigations.

Responsibilities

• Design, build, and evaluate system-level controls for agent actions in real environments.
• Collaborate with Codex harness engineering to productionize AI controls.
• Red-team end-to-end agentic systems to test data exfiltration and unsafe tool use.
• Improve safety-productivity tradeoffs by reducing missed harms and latency.

Skills

System security

Threat modeling

Experimentation

AI alignment

Risk assessment

Tools

Sandboxing

Process isolation

Codex harness

Job description

OpenAI seeks a safety-focused researcher/engineer in San Francisco to design, implement, and test system-level controls for increasingly capable AI agents in real-world environments. The role emphasizes building practical mitigations, rigorous threat models, and evaluation infrastructure, with a strong focus on safety and alignment.

The role supports a hybrid work model, collaborating with Codex teams and other security specialists to advance responsible deployment of AI systems.