How the program works.
A six-month fellowship for Argentine researchers and students, with full-time and part-time tracks.
Overview
AISAR is a program supporting Argentine academic researchers in adopting AI Safety research lines, while giving talented students the opportunity to delve into the same field. Each scholar works on a research project in a specific AI Safety topic for 6 months, choosing between full-time and part-time tracks. Scholars receive mentorship from a professor or researcher in Argentina. The program also includes a compute fund to run experiments, and seminars with international researchers working on AI Safety.
In the 2025 edition, the program was hosted at the University of Buenos Aires (UBA), supervised by the Research Office of the UBA-CONICET Institute of Computer Science (ICC).
Scholar profile
We are looking for advanced students or recent graduates from the Master's degree in Computer Science or Data Science from Exactas-UBA.
The following will be considered positively:
- Solid academic background demonstrating deep knowledge in an area related to the scholarship topic.
- Previous experience or marked interest in research projects related to Artificial Intelligence (AI) and/or Machine Learning.
- Ability to quickly incorporate advanced technical concepts from the state of the art in AI.
- Prior knowledge or strong motivation to contribute to AI Safety, particularly in mitigating catastrophic risks from advanced autonomous AI systems.
- Technical strengths in Computer Science or Data Science that, with proper guidance, can contribute to research aimed at risk reduction in AI systems.
- Adequate English proficiency for reading academic papers and participating in seminars with international AI Safety researchers.
- Availability and commitment to maintain the chosen track, full-time (~40 hours/week) or part-time (~20 hours/week), throughout the 6 months of the program.
While the profile primarily targets advanced students or recent graduates in Computer Science, Artificial Intelligence, or Data Science, we are open to other related disciplines in exceptional cases. If this is you, please include a note explaining why your profile is suitable for this call.
Research topics
Research projects are defined collaboratively by mentors and AI Safety Research Advisors before the scholar selection process, ensuring alignment with AI Safety agendas, including:
- Interpretability. Methods to understand and explain the internal and external behavior of AI models: reverse-engineering architecture, analyzing implicit goals or values, evaluating reasoning faithfulness, studying inductive biases.
- Evaluations. Behavioral and mechanistic tests for detecting dangerous capabilities, harmful propensities, and the robustness of safety measures, creating evidence-based "gates" that determine if and how a model can be scaled or deployed.
- Scalable oversight and control. Techniques like task decomposition, process supervision, debate, iterated amplification, and adversarial control evaluations that let humans (or weaker AIs) monitor and constrain much more capable models.
- Governance. Policies, regulations, institutions, and coordination mechanisms to avoid or mitigate harmful dynamics in advanced AI development (multipolar traps, unilateralist curse, proliferation of dangerous capabilities).
- AI agency. Mathematical theory of agency, optimization, corrigibility, and embedded decision-making, supporting alignment strategies valid for radically superhuman agents.
- Safety by design. AI systems developed to operate safely, with formal and verifiable guarantees of avoiding harmful behavior.
We also use as a reference the agendas and projects defined in other AI Safety programs and funding institutions:
Frequently asked questions
Scholars receive a monthly stipend based on their track: $1,600 USD/month for full-time scholars (~40 hours/week) and $800 USD/month for part-time scholars (~20 hours/week), for the 6 months of the program.
Scholars receive a monthly compute fund: $1,000 USD/month for full-time and $500 USD/month for part-time, reflecting full-time scholars' greater capacity to run experiments. As a reference, in 2025 these funds were spent roughly on AI APIs and platform subscriptions (~66%), GPU compute (~19%), and development tools and other (~15%).
Mentors are paid $50 USD/hour for the 6 months of the program, with a time commitment ranging from 2 to 8 hours per week depending on their involvement. The minimum of 2 hours per week covers direct project supervision: meetings with their scholar and feedback on research progress. Mentors with higher commitments dedicate additional hours to their own AI Safety upskilling: engaging with the literature, attending seminars, writing research and funding proposals, and collaborating with the Research Advisors.
The program team includes two key support roles:
- Research Advisors: work with mentors to define and refine their AI Safety research agendas, connect them with international researchers in the area, give feedback on funding proposals, and support their growth as research leads.
- Research Managers: hold weekly meetings with scholars and biweekly meetings with mentors, and run the program's weekly reading group.
The program lasts 6 months.
Scholars choose between two tracks: full-time (~40 hours/week) or part-time (~20 hours/week). We aim for roughly half of scholars to participate in each track, depending on availability and academic situation. The part-time track is designed so active students can combine the program with their coursework, and even use the project as their thesis. The full-time track targets recent graduates and others who can commit exclusively to the program.
Exact dates for the 2026 edition will be announced soon. As a reference, the program's structure includes:
- Application window opens and closes (~2 weeks).
- Scholar selection process (~1 month).
- Acceptance notifications and program start.
- Month 1: research plan submission.
- Month 3: mid-term report submission.
- Month 6: final workshop with scholar presentations and final report submission.
The number of scholars for the 2026 edition will depend on confirmed funding, with a maximum of 16. As a reference, in 2025 we selected 6 scholars from 51 applicants (12% acceptance rate). The process is based on:
- Analysis of candidates' CV and cover letter: degree progress, GPA, previous work or research experience, knowledge of Machine Learning, knowledge of AI Safety agendas, knowledge of English, awards received.
- Remote Python programming exam (90 minutes).
- Remote AI Safety papers reading exam (120 minutes).
- Interview with program organizers (if passing first filter).
- Matching process between scholars and mentors (if passing final round).
During the program, the following events take place:
- Program kickoff meeting.
- Integration and networking activities.
- Project follow-up meetings with mentors and research managers.
- Seminars with international researchers.
- Seminar on funding AI Safety projects (LTFF, Coefficient Giving).
- Internal workshop with scholar presentations.
- Final presentation of results.
Within the framework of the program, scholars are required to:
- Complete a tentative research plan within the first month.
- Complete the Blue Dot AI Safety Fundamentals course within the first month.
- Submit a mid-term progress report.
- Make a final presentation at an internal program workshop, showing the results obtained for the project.
- Submit a final program report.
No, you don't need to have a project in mind to apply.