Alice (Formerly ActiveFence) Benefits Overview

Adversarially test and harden Generative AI models against cyber-enabled CBRNE threats by designing and executing red-team prompts, jailbreaks, multi-turn attack scenarios, and model evaluations. Create realistic cyber-CBRNE attack scenarios involving ICS/OT and critical infrastructure, document failure modes, reproduce findings, and provide actionable mitigations. Develop test suites, adversarial datasets, evaluation rubrics, and risk taxonomies while keeping current on GenAI attack techniques and CBRNE security risks.

Perform adversarial red-teaming and risk evaluations of generative AI to identify biological dual-use risks, translate regulatory standards into safety feedback, and update benchmarks based on emerging biotech threats.

Assess and strengthen safety guardrails for Generative AI handling chemical information via adversarial red-teaming, taxonomy audits, model evaluation, and regulatory alignment. Identify dual-use risks, prevent material enablement of chemical harms, maintain knowledge of chemical conventions and controls, and communicate technical findings to safety teams.