Teams at Alice (Formerly ActiveFence)

Adversarially test and harden generative AI models against cyber-enabled CBRNE threats by designing prompts, jailbreaks, multi-turn attack scenarios, and evaluation rubrics; develop red-team test suites, document failure modes, reproduce findings, and provide actionable mitigations to AI safety and model-alignment teams.

Perform adversarial red-teaming and risk evaluations of generative AI to identify biological dual-use risks, translate biosafety/regulatory standards into model safety requirements, and advise AI safety teams while monitoring emerging biosecurity threats and updating benchmarks.

Evaluate and strengthen Generative AI safety guardrails for chemical content through adversarial red-teaming, taxonomy audits, prompt research, and risk reporting aligned with chemical regulations and standards.