AI Safety Argentina

AI Safety Argentina (AISAR) is a research scholarship initiative hosted at the UBA-CONICET Institute of Computer Science (ICC) in Buenos Aires, Argentina. Its mission is to promote AI safety research in Argentina by connecting talented local students with established international researchers and incubating the next generation of AI safety contributors in Latin America. The program runs for six months. Scholars commit approximately 20 hours per week to an AI safety research project, receiving a monthly stipend of $600 USD and up to $500 USD per month in compute resources. The cohort participates in a structured program including kickoff meetings, regular mentor check-ins, seminars with international researchers, a funding seminar, and a final internal workshop with presentations. Research areas include interpretability, evaluations, scalable oversight and control, AI agency, and safety-by-design. The first cohort application period ran in May 2025, with the program launching on June 30, 2025 and concluding in December 2025, supporting up to six scholars. Open Philanthropy recommended a grant of $77,000 to support this first edition. The program is organized by Iván Arcuschin (independent AI safety researcher, PhD in Computer Science from UBA, MATS fellow focused on unfaithful chain-of-thought in LLMs) and Agustín Martinez Suñé (postdoctoral research associate at the University of Oxford's OXCAV group, PhD in Computer Science from UBA, former PIBBSS fellow). The advisory board includes Sebastián Uchitel (UBA professor, IEEE TSE editor-in-chief), Luciana Ferrer (ICC-CONICET researcher, Stanford PhD), Ryan Kidd (MATS co-director), James Fox (Schmidt Sciences), and Nora Anmman (ARIA technical specialist, PIBBSS co-founder). AISAR is closely related to BAISH (Buenos Aires AI Safety Hub), a community group at UBA with 200+ members that runs introductory courses, research workshops, and a paper-reading club. Multiple BAISH community members have advanced through the AISAR scholarship. Together they form the core of the AI safety community in Argentina.