About the job
Join the Applied Foundations team at OpenAI, where we are committed to not only revolutionizing technology but also safeguarding it against a wide array of adversarial threats. Our mission is to uphold the integrity of our platforms as they expand. We are at the forefront of defending against financial abuse, scaled attacks, and other forms of misuse that could compromise user experience or operational stability.
The Integrity pillar within Applied Foundations is charged with overseeing the systems that identify and address harmful activities on OpenAI’s platforms. As we enhance our capabilities to tackle some of the most pressing usage harms, we are seeking experienced data scientists to robustly measure the prevalence of these issues and evaluate the effectiveness of our responses.
About the Role
We are looking for seasoned trust and safety data scientists to enhance, operationalize, and monitor measurements for intricate actor- and network-level harms. In this role, you will take ownership of measurement and metrics across various established harm verticals, including estimating the prevalence of on-platform (and occasionally off-platform) harm, as well as analyses to uncover gaps and opportunities in our responses.
This position is based in our London office and may require addressing urgent escalations outside of standard working hours. Some areas of concern may involve sensitive content, including sexual, violent, or otherwise disturbing material.
In this role, you will:
Take charge of measurement and quantitative analysis for a range of significant actor- and network-based usage harm verticals.
Design and implement AI-first methods for prevalence measurement and other safety metrics, potentially utilizing off-platform indicators or non-traditional datasets.
Develop metrics suitable for goal-setting or A/B tests when standard top-line metrics are inadequate.
Manage dashboards and metrics reporting for harm verticals.
Perform analyses and generate insights that drive improvements in review, detection, or enforcement, influencing strategic roadmaps.
Optimize LLM prompts for the purpose of measurement.
Collaborate with other safety teams to grasp key safety concerns and formulate relevant policies to support safety objectives.
Engage in additional duties as required to ensure the integrity and safety of OpenAI’s platforms.

