Jobs at Handshake

Evaluate AI models for dangerous CBRNE-related outputs, designing adversarial prompts, multi-turn attack chains, and threat models. Assess technical accuracy and real-world risk, distinguish open-source information from harmful uplift, score responses using severity rubrics, and document failures. Collaborate with AI researchers and policy teams to improve model safeguards and evaluation frameworks while following strict ethical and operational security protocols.

Evaluate AI requests and responses involving violence, weapons, threats, and dark fiction. Apply policy and conversation context to distinguish legitimate fictional, educational, historical, or defensive content from real-world harm enablement or credible threats. Write concise rationales, create adversarial prompts, identify policy gaps, participate in calibration, and maintain consistent judgment while reviewing graphic and sensitive material.