CBAI is a Cambridge, MA-based 501(c)(3) nonprofit that runs research fellowships and technical bootcamps to grow the pipeline of AI safety researchers, and fiscally sponsors student AI safety groups at Harvard and MIT.
CBAI is a Cambridge, MA-based 501(c)(3) nonprofit that runs research fellowships and technical bootcamps to grow the pipeline of AI safety researchers, and fiscally sponsors student AI safety groups at Harvard and MIT.
People
Updated 05/18/26Director of Programs
Co-founder and Executive Director
Director of Operations
Board Member
Board Member
Research Manager
Research Manager
Research Manager
Funding Details
Updated 05/18/26- Annual Budget
- $1,128,480
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 05/18/26The Cambridge Boston Alignment Initiative (CBAI) is a 501(c)(3) public charity based in Cambridge, Massachusetts (EIN: 92-1463153). Founded in 2022 by Kuhan Jeyapragasan, Trevor Levin, and colleagues, CBAI’s mission is to advance research and education so that society navigates a safe and beneficial transition to advanced AI systems. CBAI runs intensive, fully funded research fellowship programs in Harvard Square, including a ten‑week Spring Research Fellowship and a nine‑week Summer Research Fellowship in AI safety. Recent summer cohorts select up to around 30 fellows working on technical AI safety and AI governance. Fellows receive a $10,000 stipend over nine weeks, housing support for non‑local participants, weekday lunches and dinners, 24/7 office access in Harvard Square, and up to $10,000 in compute and conference support. Fellows work closely with external mentors and in‑house research managers and attend weekly workshops and speaker events with researchers from institutions such as Harvard, MIT, Northeastern, Goodfire, AVERI, the Institute for Progress, and other leading AI safety organizations. CBAI also operates CAMBRIA (Cambridge Bootcamp for Research in Interpretability and Alignment), a three‑week ML upskilling bootcamp based on the ARENA curriculum. CAMBRIA runs multiple times per year in Cambridge and other cities, covering transformers, interpretability, reinforcement learning, and alignment for participants with prior experience in Python, multivariable calculus, and linear algebra. In addition, CBAI has launched a fully funded, nine‑week AIxBiosecurity Research Fellowship in Cambridge, focused on research at the intersection of advanced AI and biosecurity, and continues to expand its fellowship offerings across Spring, Summer, and AIxBio cycles. As a Model A fiscal sponsor, CBAI supports student‑led AI safety organizations at Harvard and MIT, including the Harvard AI Safety Student Teams (AISST) and MIT AI Alignment (MAIA). This sponsorship provides office space, funding, and operational infrastructure that enables these groups to run bootcamps, workshops, and research programs. CBAI’s work is supported by philanthropic funders such as Coefficient Giving (formerly Open Philanthropy) and other donors. The organization is led by a growing professional team including a Director of Programs, Director of Operations, multiple research managers, and operations staff, with a board of directors that includes co‑founder Trevor Levin and alignment researcher Samuel Marks.
Theory of Change
Updated 05/18/26CBAI's theory of change centers on the idea that a primary bottleneck to reducing existential risk from advanced AI is the shortage of talented, well-trained researchers working on alignment and safety. By running research fellowships and technical bootcamps in the Boston/Cambridge area — home to Harvard, MIT, and Northeastern — CBAI aims to accelerate the career trajectories of promising students and early-career researchers, exposing them to mentorship from leading safety researchers and giving them hands-on research experience. Supporting student organizations at Harvard and MIT through fiscal sponsorship builds durable institutional communities that recruit and train new cohorts of safety-minded researchers each year. Taken together, these programs are intended to meaningfully increase the quantity and quality of people capable of producing high-impact AI safety research, thereby improving the field's ability to make AI development go well for humanity.
Grants Received
Updated 05/18/26Projects
Updated 05/18/26A three‑week ML upskilling bootcamp for aspiring technical AI safety researchers, based on the ARENA curriculum and run by CBAI, covering transformers, mechanistic interpretability, reinforcement learning, and alignment in Cambridge and other cities.
A fully funded nine‑week research fellowship in Cambridge, MA at the intersection of advanced AI and biosecurity, supporting fellows to work with mentors on AI‑enabled bio risk, evaluations, and related governance questions.
Discussion
No comments yet. Be the first to share your thoughts.