Model Policy & Agentic Safety Lead

OpenAI, Inc.; San Francisco (CA); Ibrido; USD 180.000 - 240.000; Tempo pieno

Vantaggi offerti da questo lavoro

Relocation support

Hybrid work model

Descrizione del lavoro

OpenAI is seeking a Safety Systems role focused on policy and model misalignment as AI models become more capable and autonomous. You will investigate real-world harms, translate insights into behavioral policies, and develop safeguards.

The role is based in San Francisco with relocation support and a hybrid model: three days in the office per week with optional work from home on Thursdays and Fridays. You will work with research, engineering, security, product, and policy teams to shape model

Competenze

• Background in AI agent safety, privacy, security, or adjacent fields.
• Strong interest in AI alignment and understanding of misaligned model behavior.
• Sufficient technical fluency to work with evaluation data and training data.
• Hands-on experience with model data and evaluation results.

Mansioni

• Identify vulnerabilities as models interact with tools, data, and external systems.
• Develop threat models and empirical frameworks for harmful outcomes from misalignment.
• Build frameworks to understand harmful outcomes from misalignment.
• Identify underlying behaviors and system conditions driving those outcomes.
• Turn findings into policy frameworks, evaluation criteria, and safeguards.
• Develop human data campaigns and gold sets for measurement and evaluation.
• Collaborate with research, engineering, security, and product teams to balance safety, utility, and risk.
• Inform deployment decisions and OpenAI's safety approach.
• Build monitoring approaches to detect regressions and emerging risks after deployment.

Conoscenze

AI safety

Model evaluation

Policy development

Threat modeling

Data analysis

Descrizione del lavoro

OpenAI is seeking a Safety Systems role focused on policy and model misalignment as AI models become more capable and autonomous. You will investigate real-world harms, translate insights into behavioral policies, and develop safeguards.

The role is based in San Francisco with relocation support and a hybrid model: three days in the office per week with optional work from home on Thursdays and Fridays. You will work with research, engineering, security, product, and policy teams to shape model