Researcher, Frontier Risk Mitigations

Written by the T&S Network from OpenAI's posting.

Frontier AI safety mitigation research within OpenAI's Preparedness team, which measures and predicts capabilities of frontier AI systems and builds safeguards against misuse and misalignment before deployment. The work sits in the team's mitigation track, aimed at extreme threats the company expects its deployed models could pose.

• AI misuse and jailbreaks

What you'd do

• Identify emerging AI safety risks and develop new methods to explore and mitigate them
• Build and continuously refine evaluations that assess the extent of these risks
• Work with internal or external domain experts on risk evaluations where needed
• Set research directions and strategies to make AI systems safer, more aligned, and more robust
• Contribute to AI safety best-practice guidelines for OpenAI and the wider industry
• Design and evaluate red-teaming pipelines to test end-to-end robustness of safety systems

What it asks for

• Doctoral or other advanced degree in a computing, machine learning, or adjacent technical field
• Four or more years in research engineering roles
• Working fluency in Python or a comparable programming language
• Experience working in environments involving large-scale AI systems

Terms in this role

Red teaming.

Testing a system, often an AI model, by deliberately trying to make it fail or misbehave, so problems are found and fixed first.

RLHF.

Reinforcement learning from human feedback: training an AI model on people's ratings of its answers.

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on OpenAInorth_east before you apply. If we got something wrong, report this listing.