Abuse Investigator

About the Team OpenAI’s mission is to ensure that artificial general intelligence benefits all of humanity. We believe this requires real-world deployment, continuous learning, and rapid iteration to improve both safety and utility. The Intelligence & Investigations team supports this mission by identifying, investigating, and disrupting malicious uses of OpenAI products. We work at the intersection of investigations, analytics, policy, engineering, and trust & safety to understand emerging threats, develop scalable detection strategies, and improve platform defenses. Our team combines investigative expertise with technical analysis to uncover novel abuse patterns, inform product and policy decisions, and help build durable safeguards across OpenAI’s products. About the Role As an Abuse Investigator on the Intelligence & Investigations team, you will investigate harmful activity on OpenAI’s platform with a particular focus on violent threats, coordinated abuse, and other high-risk threats to public safety. You will analyze complex datasets, identify abuse signals, conduct investigations, and partner closely with policy, legal, engineering, and integrity teams to drive enforcement actions and improve detection capabilities. This role is highly analytical and requires strong technical and investigation skills. Successful candidates are comfortable working directly with large datasets, using SQL to uncover patterns and trends, and leveraging Python or other tools like Codex to automate workflows, analyze data, and support investigations at scale. You will be expected to combine investigative judgment with data-driven analysis to identify emerging threats and develop proactive mitigation strategies. The role is based in New York City and follows a hybrid in office work model. Investigations may involve sensitive content, including violent, extremist, sexual, or child safety-related material. Candidates should be comfortable operating in high-stakes environments and managing time-sensitive escalations. In This Role, You Will - Investigate reports, leads, and escalations involving violent threats, coordinated abuse, and other threats to public safety. - Use SQL and Codex to query large datasets, analyze user behavior, identify suspicious activity, and uncover novel abuse patterns. - Utilize Python and Codex to automate investigative workflows, process and analyze data, develop abuse detection methodologies, and improve operational efficiency. - Develop and refine abuse signals, behavioral indicators, and tracking methodologies that enable scalable detection and enforcement. - Partner closely with engineering teams to design and improve internal tooling, data pipelines, reporting capabilities, and investigations. Use metrics to evaluate automated workflow efficiency. - Conduct rapid triage and investigation of urgent escalations, exercising sound judgment in ambiguous situations. - Communicate investigative findings and recommendations to stakeholders across Policy, Legal, Product, Engineering, and Leadership teams. - Translate investigative insights into long-term product, policy, and safety improvements. - Contribute to an on-call rotation supporting time-sensitive investigations and threat assessments. You Might Thrive in This Role If You - Have 5+ years of experience conducting investigations involving violent actors, terrorism-related activity, trust & safety, law enforcement investigations, intelligence analysis, or related fields (ideally with a public safety component) - Possess strong SQL skills and experience using data analysis to support investigations, identify trends, and answer complex investigative questions. - Have practical experience using Python for data analysis, workflow automation, investigative tooling, or operational analytics. - Have experience developing scalable processes, detection methodologies, or automated approaches to investigations. - Can independently work through complex problems and clearly communicate findings and recommendations. - Have experience collaborating with technical and non-technical stakeholders in fast-moving environments. - Have experience presenting analytical findings through reports, briefings, dashboards, or executive-level communications. - Have worked in technology, trust & safety, cybersecurity, intelligence, law enforcement, fraud, risk, or investigative environments. - Enjoy training and/or mentoring others on technical tooling and workflows. Preferred Qualifications - Experience conducting investigations involving violent threats, online abuse, extremist activity, or other threat actor behavior. - Familiarity with data analytics tools and investigative methodologies used in technology, intelligence, or trust & safety environments. - Experience partnering with engineering teams to improve detection systems, data quality, or investigative tooling. - Understanding of emerging risks associated with AI systems, online platforms, and

5+ years of experience conducting investigations involving violent actors, terrorism-related activity, trust & safety, law enforcement investigations, intelligence analysis, or related fields (ideally with a public safety component). Strong SQL skills and experience using data analysis to support investigations, identify trends, and answer complex investigative questions. Practical experience using Python for data analysis, workflow automation, investigative tooling, or operational analytics. Experience developing scalable processes, detection methodologies, or automated approaches to investigations. Ability to independently work through complex problems and clearly communicate findings and recommendations. Experience collaborating with technical and non-technical stakeholders in fast-moving environments. Experience presenting analytical findings through reports, briefings, dashboards, or executive-level communications. Experience working in technology, trust & safety, cybersecurity, intelligence, law enforcement, fraud, risk, or investigative environments. Enjoy training and/or mentoring others on technical tooling and workflows.