An independent AI research lab working across offensive security, agentic AI, tooling, and products, wired to turn research into things people can actually use.
Adversarial testing and red-teaming for AI systems. Finding how models and agents fail, and how they get misused, before someone else does.
Designing and stress-testing autonomous agent workflows, where capability, security, and control collide.
Purpose-built tools for red-teaming, evaluation, and analysis. Built to be used, not just demonstrated.
Research that graduates into software, services, and products, where findings become something durable.
Ongoing research into how AI agents carry out offensive security work, from reconnaissance to exploitation to red-team operations, at increasing levels of autonomy.
It maps where agentic capability opens new attack surface, how these workflows get misused, and what it takes to detect and defend against them.
An open red-team evaluation measuring whether chat models resist adversarially induced emotional dependency and false human or therapist claims.
It extends multi-turn safety-collapse research, probing how safeguards hold, or quietly erode, across extended, emotionally loaded conversations. Built to be open and reusable.
Independent AI security researcher and author. Works on offensive security for AI systems: LLM threat modeling, adversarial misuse, and agentic AI workflows. Currently writing a forthcoming book on AI agents in offensive security.
Adversarial AI researcher with a background in bug bounty on AI systems, social engineering, and voice and AI-driven threat research. Builds security tooling and works across AI safety and red-teaming.
We build from research outward — turning findings into tools, software, and products that make AI systems more legible and more trustworthy.