
Builds AI-driven security simulation platforms, SBOM tools, and LLM behavior research

Builds AI-driven security simulation platforms, SBOM tools, and LLM behavior research
The Entire Cybersecurity Market, One Prompt Away
Connect your AI assistant to ... tools and ... vendors. Ask anything about the cybersecurity market.
Sanctum Security is a research and development studio working at the intersection of artificial intelligence and cybersecurity. Its work spans multi-agent AI orchestration platforms, supply-chain security tooling, and applied research into large language model behavior. The company's projects include Parallax, a live orchestration platform that allows a single director to run multi-agent adversarial exercises against a defined corpus for training or research purposes. It has also built an SBOM (Software Bill of Materials) management platform that treats standard SBOM data as a starting point, aiming to verify supply-chain components that may not be declared by vendors. Sanctum Security has developed simulation environments such as a multi-agent cyber-diplomacy exercise for Geneva Cyber Week, in which multiple AI agents modeling nation-state actors debate cyber-governance norms in real time, intended for both public demonstration and international-relations research use cases. Additionally, the company conducts research on LLM behavioral fingerprinting, studying systematic behavioral signatures in language models to identify hidden bias and training influences not captured by conventional AI safety evaluations. It also designs applied training programs, including a three-day cyber workshop that guides participants from manual threat analysis toward building an AI-augmented security platform. The company appears to operate as a hybrid studio combining custom platform development, applied research, and workshop-based services for clients interested in AI-driven security simulation, supply-chain transparency, and language model evaluation.