Luthien is developing AI Control, as proposed in "AI Control: Improving Safety Despite Intentional Subversion" (2023), into production-ready solutions that can be deployed in real-world AI systems.
Our flagship project, the Luthien Proxy, is an OpenAI-compatible LLM gateway that brings Redwood-style AI Control to production agentic deployments. It provides policy orchestration, decision logic, and full observability for your AI systems.
Based in Seattle, Luthien is the only AI Safety organization headquartered in the Pacific Northwest. Our secondary mission is to establish a robust AI Safety presence in Seattle, creating a community of researchers and engineers dedicated to ensuring AI systems remain safe and aligned with human values.
Luthien was founded by Jai Dhyani, a former Meta ML Software Engineer, MATS alumnus, and co-author on METR's "RE-Bench: Evaluating frontier AI R&D capabilities of language model agents against human experts" (2024).
Currently operating as a minimal organization, Luthien is looking to grow quickly. We exist thanks to generous small donors and a seed grant from the AI Safety Tactical Opportunities Fund. Your donations will directly accelerate the development of practical AI Control solutions.
Donate to help us expand our work on production-ready AI Control. Every contribution makes a difference.
EIN: 99-3421309 (charitable status pending)
Contact us to learn more about our work or to discuss partnership opportunities.