Powerful AI systems are arriving faster than we can fully understand or reliably control. We fund the research that keeps humanity in the loop.
We provide the infrastructure, funding, and support that AI safety researchers need to do their best work — without compromise.
We identify and address funding gaps and operational barriers that slow down critical safety research, ensuring researchers can focus on what matters most.
Beyond funding, we offer engineering resources, compute infrastructure, and domain expertise to accelerate research outcomes and reduce operational overhead.
We provide multi-year, consistent funding commitments so researchers can pursue deep, longitudinal work rather than chasing short-term grants.
We track and report our impact clearly, holding ourselves accountable to donors, researchers, and the broader public with honest assessments of progress.
Advanced AI systems will soon operate across national security, infrastructure, finance, medicine, and policy. The stakes demand rigorous, independent safety research.
Interpretability research to understand how and why AI systems arrive at their outputs — making the black box more transparent and verifiable.
Building isolated, rigorous testing environments where AI systems can be evaluated safely before deployment in high-stakes real-world contexts.
Detecting and preventing AI systems from developing deceptive behaviors — ensuring they remain transparent and honest about their capabilities and actions.
We bring a principled and evidence-driven approach to one of the most consequential challenges of our time.
We don't favor any single alignment theory. We fund the most promising research regardless of its origin or approach.
Every funding decision is grounded in careful analysis and measurable outcomes — not hype or speculation.
We actively explore overlooked research directions that larger institutions miss or deprioritize.
No corporate revenue, no conflicts of interest. Our only obligation is to the mission of safe and beneficial AI.
We have the infrastructure and expertise to support real empirical testing — not just theoretical modeling.
We work alongside researchers, institutions, and peer foundations to maximize collective impact across the field.
Your support funds the urgent research needed before advanced AI systems exceed our ability to understand or control them.