SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
OpenAI's Preparedness team is seeking exceptional researchers to develop novel safety mitigations for frontier AI systems. As models transition from assistive tools to autonomous agents capable of real-world planning and execution, mitigating associated risks is critical to safe deployment.
In this role, you will work on three core pillars: (1) Measurement—monitoring and predicting evolving capabilities of frontier AI systems; (2) Mitigation—maintaining safeguards, alignment tools, and security measures to address extreme threats; and (3) Coordination—setting mitigation targets and partnering across teams to achieve them.
Key responsibilities include identifying emerging AI safety risks and developing methodologies to explore and mitigate their impact. You will build and continuously refine evaluations to assess risk extent, collaborating with domain experts (internal and external) across misalignment, cybersecurity, and biology. You will set research directions and strategies to make AI systems safer, more aligned, and robust; contribute to industry-wide AI safety best practices; and design red-teaming pipelines to examine end-to-end robustness of safety systems.
This is urgent, fast-paced work with far-reaching implications for the company and society. You will play a critical role in defining what a safe AI system should look like at OpenAI and contribute directly to the mission of building and deploying safe AGI.
Ideal candidates bring 2+ years of AI safety experience (RLHF, human-AI collaboration, interpretability, control), a Ph.D. in computer science or machine learning, 4+ years of research engineering experience, and proficiency in Python or similar languages. You should be deeply aligned with OpenAI's mission and charter, excited about long-term AI safety, and comfortable applying methods from interpretability, robustness, alignment, and control domains to real-world problems at scale.
