Researcher, Frontier Risk Mitigations

About The Team

The Preparedness team focuses on addressing significant risks associated with highly capable AI models that transition from assisting humans to independently planning and acting in real-world environments. Their efforts involve monitoring AI capabilities, ensuring safeguards against misuse and alignment failures, and collaborating across teams to maintain OpenAI’s preparedness framework.

Role Overview

This position is aimed at outstanding researchers dedicated to innovating safety mitigation strategies for frontier AI systems. Responsibilities include developing new safety techniques inspired by interpretability, control theory, and alignment research to ensure safe deployment of OpenAI’s models. The role involves defining the future shape of safe AI systems and critically contributing to OpenAI’s mission of building safe Artificial General Intelligence (AGI).

Responsibilities

• Identify and analyze emerging AI safety risks, developing innovative approaches to evaluate and reduce their impact.
• Create and continuously improve evaluation methods to assess risk levels, collaborating with domain experts as necessary.
• Guide the research agenda and strategic approach towards enhancing AI safety, alignment, and robustness.
• Participate in crafting industry and organizational best practices for AI safety standards.
• Design and review red-teaming processes to rigorously test safety systems and highlight potential improvements.

Ideal Candidate Attributes

• Strong alignment with OpenAI's mission to develop safe and beneficial AGI.
• Passion for long-term AI safety and deep understanding of technical pathways toward safe AGI.
• Practical experience applying interpretability, robustness, alignment, and control techniques to AI systems.
• Minimum of 2 years applied experience in AI safety domains including reinforcement learning from human feedback (RLHF), human-AI collaboration, or control.
• Ph.D. or an advanced degree in computer science, machine learning, or related disciplines.
• Ability to work with large-scale AI architectures effectively.
• At least 4 years of research engineering experience with proficiency in Python or comparable programming languages.

Equal Opportunity and Additional Information

OpenAI is dedicated to fostering diversity and inclusion, ensuring equal opportunity regardless of background or protected characteristics. Background checks are conducted in compliance with applicable laws. Reasonable accommodations are available for applicants with disabilities upon request.

Compensation

The salary range for this role is between $295,000 and $445,000 per year.