AFFINE (Agent Foundations FIeld NEtwork) runs intensive superintelligence alignment seminars and fellowships to upskill promising newcomers in agent foundations and AI alignment research.
AFFINE (Agent Foundations FIeld NEtwork) runs intensive superintelligence alignment seminars and fellowships to upskill promising newcomers in agent foundations and AI alignment research.
People
Updated 03/19/26Founder
Funding Details
Updated 03/19/26- Annual Budget
- -
- Current Runway
- -
- Funding Goal
- $1,616,120
- Funding Raised to Date
- -
Org Details
Updated 03/19/26AFFINE (Agent Foundations FIeld NEtwork) is an AI safety field-building organization focused on providing intensive upskilling in superintelligence alignment research. The organization was founded by Mateusz Baginski, a cognitive science graduate (BSc, MSc) and 2024 PIBBSS Fellow who transitioned into theoretical agent foundations research. AFFINE's flagship program is a one-month Superintelligence Alignment Seminar held at Hostacov Chateau near Prague, Czechia. The seminar accepts up to 30 participants who engage with foundational materials, think deeply about hard problems, and solidify their understanding through peer teaching and collaboration with world-class mentors. The program is fully funded for participants, covering accommodation and meals, with budget available for travel assistance. The seminar's mentor roster includes leading researchers across AI alignment: Abram Demski, Ramana Kumar (DeepMind), Steve Byrnes (Astera Institute), Kaj Sotala, Kaarel Hanni, Cole Wyeth, Aram Ebtekar, Elliot Thornley, Linda Linsefors, and Paul 'Lorxus' Rapoport. Two full-time on-site mentors, Ouro (ex-Orthogonal) and Jonas Hallgren (Equilibria Network), provide continuous guidance throughout the month. The AFFINE Fellowship extends the seminar into a year-long program for the approximately 10 most promising participants. Selection is based on collaborative excellence, seeking participants who help others learn, integrate across disciplines, and build rather than hoard knowledge. The goal extends beyond producing individual researchers to creating a cohesive network that continues collaborating after the intensive month ends. AFFINE ran its first seminar cohort from April 28 to May 28, 2025, and announced its second cohort for the same dates in 2026. The organization received $244,000 in grants from the Survival and Flourishing Fund's 2025 S-Process round, funded by Jaan Tallinn, split between Algorithm Design ($79,000) and Technical Research ($165,000). It also raised funds through Manifund for its Ambitious AI Alignment Seminar project. AFFINE operates under fiscal sponsorship from Ashgro, Inc., a 501(c)(3) public charity that provides fiscal sponsorship to AI safety projects.
Theory of Change
Updated 03/19/26AFFINE's theory of change centers on addressing a bottleneck in AI alignment research: the shortage of researchers with deep understanding of foundational alignment problems. By providing an intensive month-long immersion in superintelligence alignment theory, guided by leading mentors from organizations like MIRI, DeepMind, and Astera Institute, AFFINE aims to rapidly upskill promising newcomers. The subsequent year-long fellowship builds a persistent collaborative network among the most capable participants, creating lasting research relationships. This approach seeks to grow the pool of researchers equipped to work on agent foundations and alignment theory, areas considered important but neglected, ultimately contributing to the mitigation of AI existential risk through better theoretical understanding of agency, decision theory, and embedded agents.
Grants Received– no grants recorded
Updated 03/19/26Projects
Updated 03/19/26AFFINE is an AI safety research training network that runs an intensive month-long superintelligence alignment seminar and year-long fellowship, with Algorithm Design as one of its core funded research focus areas.
AFFINE (Agent Foundations FIeld NEtwork) runs intensive educational programs and supports technical research aimed at solving the superintelligence alignment problem, with a focus on agent foundations.
Discussion
No comments yet. Be the first to share your thoughts.