You will analyze how users think, feel, and behave in interaction with AI systems,especially in high-risk contexts such as self-harm, manipulation, coercion, and influence,and translate these insights into decision-ready risk assessments, mitigation strategies, and product guidance.

This role bridges clinical/behavioral expertise and intelligence analysis, turning psychological signals and patterns into structured judgments, early indicators, and actionable recommendations.

You will partner closely with investigators, engineers, policy, and trust & safety teams to shape how we understand and mitigate potential risks in human-AI interactions.

Key responsibilities include:

• Developing insights into how AI systems are used in complex or high-risk situations (e.g., self-harm, suicidal ideation, substance-use escalation, and threats of violence), identifying recurring patterns and emerging trends that help guide product, safety, and policy decisions.

• Synthesizing behavioral, psychological, and intelligence signals into clear narratives about user needs, system dynamics, and potential areas of risk or vulnerability.

• Producing decision-ready briefs and assessments that inform product, safety, and policy decisions.

• Developing and refining behavioral risk frameworks, taxonomies, and indicators (e.g., severity models, escalation pathways, psychological harm categories).

• Identifying early indicators of emerging issues and assessing whether observed patterns represent meaningful safety concerns, helping prioritize and inform appropriate mitigations.

• Assessing the effectiveness of mitigations,such as product changes, safeguards, and guidance,using behavioral evidence and real-world outcomes.

• Contributing to incident reviews and post-incident analysis by bringing a behavioral perspective to root cause analysis and prevention.

• Bridging research and operations, translating academic and clinical literature into practical safeguards, policies, and product decisions.

You might thrive in this role if you:

• Bring 5+ years in forensic, clinical, trust and safety, or applied academic settings assessing risk of violence, self-harm, or addiction, with strong mixed-methods research skills.

• Have familiarity with AI systems, language models, or human-AI interaction dynamics, and are interested in applying psychological expertise to emerging AI risks (experience working on AI safety, trust & safety, or related domains is a plus).

• Can translate human behavior into structured intelligence, connecting individual cases to system-level patterns and risks.

• Are comfortable working across qualitative and quantitative inputs, including casework, interaction data, research literature, and metrics.

• Have experience designing or using risk frameworks, taxonomies, or evaluation methods to structure ambiguity.

• Communicate clearly across disciplines, turning complex behavioral insights into concise, actionable recommendations.

• Thrive in fast-moving, ambiguous environments, and can prioritize effectively under uncertainty.

XML job scraping automation by YubHub