What AI Safety Camp does
AI Safety Camp (“AISC”) is an AI-safety field-building organization that runs project-based research programs to help people develop and execute concrete AI-safety work in teams. The core mechanism is team formation around participant-proposed or participant-selected projects: project leads submit proposals, AISC provides feedback, accepted projects are posted publicly for team recruitment, and then each team collaborates part-time for roughly three months with an expectation of substantial weekly participation. AISC positions its work as accelerating progress on AI safety by “learning by doing,” aiming to produce tangible research outputs and to connect participants with longer-term collaborators and career paths in AI safety. It explicitly focuses on both (1) work intended to reduce existential or other large-scale risks from advanced AI and (2) ensuring that projects do not primarily accelerate capabilities without commensurate safety value.
In addition to its ongoing camp format, AISC has also run a separate “research incubator” style program (a 16-day virtual incubator in August 2026) aimed at helping participants clarify research questions, scope a research direction, and continue into the next AI Safety Camp where desired. Operationally, AISC is run by a small group of organizers who onboard and support project leads, coordinate the application and selection process, and maintain scheduling and fiscal administration.
Strategically, AISC appears to function as a talent pipeline and research collaboration hub for early-career and transitioning contributors to AI safety—especially for people outside the traditional Bay Area/London hubs—by providing structure, mentoring, and an environment where teams can carry forward after a program ends. In 2025, AISC received substantial support through the Survival and Flourishing Fund (SFF), and as of late 2026 it was fundraising for its 12th edition (AISC12) via Manifund, with stated matching from SFF-linked matching pledges.