OpenAI
Model Policy Manager, Agentic Safety
$207K–$335K
San Francisco · Hybrid · Manager · Full-time · Posted Sep 15, 2026
About the job
Investigates risks from misaligned, increasingly autonomous models and turns evidence into behavioral policies, evaluations, monitoring, and safeguards for safe deployment.
What you'll do
- Identify vulnerabilities and harmful outcomes in model interactions with tools, data, and external systems; develop threat models and safeguards.
- Translate findings into policy frameworks, evaluation criteria, human data campaigns, online measurement, and post-deployment monitoring.
- Partner across research, engineering, security, and product to inform safety decisions, system cards, and safeguards reports.
What they ask for
- Strong background in AI agent safety, privacy, security, cybersecurity, or an adjacent field.
- Demonstrated interest in AI alignment and strong understanding of technical drivers of misaligned model behavior.
- Technical fluency working directly with evaluation and training data, inspecting examples, analyzing failures, and assessing data quality.
- Ability to use empirical evidence to define precise behavioral expectations and measurable evaluation criteria.
Pay
$207K – $335K USD base salary
AI focus
Agentic AI safety and alignment: investigating long-horizon misalignment and developing policies, evaluations, monitoring, and safeguards.
Model-written summary · the original posting is authoritative