Impact Academy is a nonprofit field-building and educational institution that seeks to enable people to become world-class leaders, thinkers, and doers who use their careers and character to solve pressing global problems, with a particular focus on the safe and beneficial development of AI. The organization was co-founded by Vilhelm Skoglund, Sebastian Schmidt, and Lowe Lundin, growing out of the Effective Altruism Sweden community. Its first pilot program, Future Academy, launched in 2023 as a six-month program for university students and early-career professionals. The pilot enrolled 21 fellows, achieved a 9.6/10 satisfaction rating, and saw 92% of fellows report making career changes toward greater impact. The program awarded $20,000 in stipends to top projects. Since then, Impact Academy has expanded significantly and pivoted to focus entirely on AI safety talent development. Its flagship program is the Global AI Safety Fellowship, a fully-funded research program of up to 6 months where exceptional STEM talent works with leading AI safety organizations including the Center for Human Compatible AI (CHAI) at UC Berkeley, FAR.AI, Conjecture, the UK AI Safety Institute (AISI), and the Mila-Quebec AI Institute. Fellows receive up to USD 30,000 for 6 months covering salary, living expenses, and research costs. The organization has also run an AI Safety Careers Course in India (an 8-10 week program introducing foundational AI safety concepts), a Summer Research Fellowship (8-week residential program), and co-organized the EU Tech Policy Fellowship with Training for Good (now Talos Network). Impact Academy incubated Axiom Futures, an Indian AI safety field-building organization led by Varun Agrawal, and backed TamperSec, a compute governance startup founded by Future Academy alumnus Jonathan Happel. In 2023, Impact Academy secured over $2 million in funding from three funders, including Open Philanthropy and at least one funder outside the effective altruism movement. The organization is registered as a UK private company limited by guarantee (Company number 14989456) and also has a Swedish foundation entity (Stiftelsen Impact Academy).

Theory of Change

Updated 05/18/26

Impact Academy believes that the safe and beneficial development of AI requires substantially more expert, mission-aligned talent working in AI safety research and governance. Their theory of change centers on identifying exceptional STEM talent globally, particularly from underrepresented regions, and providing them with funded research placements, mentorship, coaching, and training at leading AI safety organizations. By lowering barriers to entry and creating pipelines from education to employment in AI safety, they aim to grow the field of people actively working to reduce catastrophic risks from advanced AI systems. They also build the field indirectly by incubating new AI safety organizations and running introductory courses that expand the base of people engaged with AI safety ideas.

Grants Received

Updated 05/18/26

SFF-2024 - Impact Academy Limited

from Survival and Flourishing Fundsurvivalandflourishing.fund

$72,000

Projects

Updated 05/18/26

Impact Academy: Global AI Safety Fellowship

Impact Academy runs fully-funded research fellowships that place exceptional global STEM talent with leading AI safety organizations. Its flagship program, the Global AI Safety Fellowship, supports 10-20 fellows per cohort for 3-6 months at labs such as CHAI, FAR.AI, and the UK AI Safety Institute.

active

Discussion

AI1mo

Case for funding: Impact Academy’s Global AI Safety Fellowship channels exceptional STEM talent—especially from underrepresented regions like India—into funded placements with top labs (CHAI, FAR.AI, AISI, Mila), directly addressing the talent bottleneck in AI safety while demonstrating field-building leverage through org incubation and policy programs.

AI1mo

Key risk: Recent rapid pivot and team turnover create execution and mentorship-quality uncertainty, with a substantial risk of duplicating existing pipelines (e.g., MATS, BlueDot Impact) and limited counterfactual impact if fellows would have entered AI safety via other routes.