Beneficial AI Foundation (BAIF)
A US nonprofit founded by Max Tegmark and Meia Chita-Tegmark to place AI safety on a solid quantitative foundation. BAIF funds research, fellowships, and university partnerships aimed at ensuring advanced AI systems remain safe and beneficial.
A US nonprofit founded by Max Tegmark and Meia Chita-Tegmark to place AI safety on a solid quantitative foundation. BAIF funds research, fellowships, and university partnerships aimed at ensuring advanced AI systems remain safe and beneficial.
People
Updated 05/18/26President
Treasurer
Chief Scientist
Product Manager
Program Director
Funding Details
Updated 05/18/26- Annual Budget
- $670,550
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 05/18/26The Beneficial AI Foundation (BAIF) is a US 501(c)(3) nonprofit organization founded in December 2022 by Max Tegmark, an MIT professor of physics and AI, and Meia Chita-Tegmark. The organization operates under the tagline "No one knows how to make tomorrow's AI safe. Let's change that," reflecting its conviction that current AI safety approaches lack the rigorous quantitative grounding needed to handle transformative AI systems. BAIF pursues three main programs. First, in-house research focuses on mechanistic interpretability (understanding how AI systems work internally, sometimes called "artificial neuroscience") and formal verification approaches to guaranteed safe AI. Staff researchers are working on deploying AI-written formally verified code rather than untrusted neural networks, with the goal of providing hard safety guarantees. Second, the Buterin Fellowship program provides PhD stipends ($40,000/year plus tuition) and postdoctoral stipends ($80,000/year) to researchers working on AI existential safety at universities in the US, UK, and Canada. This fellowship is administered in partnership with the Future of Life Institute. Third, the Quantitative AI Safety Initiative (QAISI) is a collaborative research effort co-led by prominent researchers including Yoshua Bengio (Montreal), Stuart Russell and Dawn Song (Berkeley), Clark Barrett (Stanford), and Bryan Parno (Carnegie Mellon). QAISI aims to develop quantitative safety standards for AI analogous to the failure-rate standards used in aerospace and nuclear engineering. The organization is funded by Vitalik Buterin (Ethereum co-founder) and Erik Otto. Its registered address is in Wilmington, Delaware, and its operations are centered around Cambridge, Massachusetts (the MIT area), though researchers are distributed internationally. As of the FY2024 990 filing, BAIF reported revenues of $125,000 and expenses of $670,550, with grants made to UC Berkeley ($210,238) and the Good Ventures Foundation ($150,027). The team includes approximately 14 people per LinkedIn, though the 990 reports no full-time employees (suggesting most are contractors or part-time). Max Tegmark serves as President and Treasurer; Meia Chita-Tegmark serves as Secretary.
Theory of Change
Updated 05/18/26BAIF believes that advanced AI poses existential risks, and that current safety approaches are insufficient because they lack quantitative rigor. Just as engineers quantified failure rates for jet engines and nuclear reactors to make those technologies safe, BAIF aims to develop measurable safety standards and formal guarantees for AI. By funding mechanistic interpretability research (to understand what AI systems are actually doing), formal verification work (to mathematically certify that AI-written code does what it is supposed to do), and red-teaming and audit standards (to systematically identify vulnerabilities before deployment), BAIF seeks to give policymakers and developers the tools to verify that AI systems won't cause catastrophic harm. The Buterin Fellowship grows the pipeline of researchers working on these technical problems, while QAISI coordinates leading academics toward shared quantitative standards.
Grants Received– no grants recorded
Updated 05/18/26Projects
Updated 05/18/26Fellowship program managed by BAIF that provides fully funded PhD and postdoctoral support for researchers working on AI existential safety, run in partnership with the Future of Life Institute and funded by Vitalik Buterin.
Multi‑university collaboration, co‑organized by BAIF, to place AI safety on a solid quantitative foundation by developing methods and standards that quantify the failure risks of advanced AI systems.
Open, collaborative moonshot led by BAIF with Signal and the Lean community to formally verify the Signal protocol and its Rust implementation using the Lean theorem prover and AI‑assisted tools.
Discussion
No comments yet. Be the first to share your thoughts.