BAISH Logo
BuenosAiresAISafetyHub
AboutProgramsResearchResourcesContact
ENES

Applications for our courses are open. See courses→

BAISH / About BAISH

The largest AI safety hub in Latin America

BAISH is a community of 400+ students, researchers and professionals in Buenos Aires working to make advanced AI go well. We run in-person courses, a research lab, monthly socials and one-on-one career support, so that talented people here can get into AI safety without having to leave the country to do it.

Focus areas

  • Interpretability & control
  • Evaluations & monitoring
  • Governance & strategy
Our track record↓The team↓Supported by↓

Impact

Our track record

What BAISH members have gone on to do, and what the community has produced so far.

members on WhatsApp
400+
fellowships joined
28
full-time AI safety roles
6
publications by researchers connected to BAISH
15+

Where members work full-time

6 members now work full-time in AI safety at organizations including:

  • Apart Research ↗
  • Equistamp ↗
  • 80,000 Hours ↗

Fellowships and programs

28 fellowships joined by about 21 people since 2024.

  • ML4Good 8+
  • MARS 4
  • SPAR 4
  • ARENA 2
  • Cambria 2
  • Apart Research SPS Fellowship 2
  • Formal methods × AI safety internship 2
  • CLR Summer Research Fellowship 1
  • Iliad 1
  • CORDA Democracy Fellowship 1
  • Algoverse 1

Research output

  • 3 papers at ICLR 2026 workshops by AISAR scholars
  • 1 NeurIPS paper from an Apart Research hackathon-winning team
  • PowerBench, a BlueDot-funded study born at the Apart Global South hackathon
  • 8 AI-safety-related theses in progress

Courses

  • 6 in-person cohorts since March 2026
  • 167 applications for 89 seats
  • Average recommendation score: 9.4/10

Community

  • Monthly socials with 50–80 attendees, about half of them newcomers
  • 500+ subscribers on our Luma calendar
  • 16-person Argentine delegation to EAG London 2026
  • Annual retreat for the organizing team
  • 970+ followers on Instagram, the largest AI safety account in the Spanish-speaking world

Success stories

  • “Thanks to the group and to Eitan, who backed me a lot through the application process.”
    TM

    Tobías Martin

    MIRI

    Applyathon → MARS

  • “BAISH gave me the foundation and connections to transition into AI safety research professionally.”
    GB

    Guido Bergman

    Former Courses Lead · Research Engineer, Equistamp

    Courses Lead → Algoverse → Equistamp

  • Took the intensive course in March 2026 and was accepted to ARENA two months later, where his team's project won first place. He now leads research at BAISH Labs and runs the Technical AI Safety Project.

    Tobías Bersia

    Technical AI Safety Project lead · Research Lead, BAISH Labs

    TAIS Intensive → ARENA → BAISH Labs

  • BAISH supported his trip to EAG London 2026. The connections he made there led directly to a three-month Iliad fellowship, and he is now a research fellow at BAISH Labs.

    Julián Szereszewski

    Research Fellow, BAISH Labs · Iliad alumnus

    EAG London → Iliad → BAISH Labs

  • Went through four external programs with BAISH's support, received a BlueDot career transition grant, and co-authored "Is Gemini 3 Scheming in the Wild?" with BAISH advisors.

    AW

    Alejandro Wainstock

    CLR

    SPAR → ML4Good → ARENA → CLR

  • “Thanks BAISH for putting this opportunity on my radar.”
    JC

    Juan P. Cadile

    Cambria

    BAISH community → Cambria

  • Accepted to the Cambria fellowship in June 2026 after the Technical AI Safety course, which he credits for getting in.

    GB

    Guillermo Bondonno

    Cambria

    Technical AI Safety course → Cambria

  • From local meetups to international research collaborations.

    AV

    Ana & Vicky

    ML4Good Camp

    BAISH → ML4Good Camp

The team

Directors

  • Eitán Sprejer

    Co-founding Director · Research Manager Lead, Apart Research

    Eitán is Research Manager Lead at Apart Research and a co-founding director of BAISH. He has co-authored research accepted to NeurIPS workshops, AI persuasion research, chain-of-thought monitorability research, and an ICLR 2026 workshop paper on white-box monitoring. He holds a bachelor's and master's degree in Physics from the University of Buenos Aires, was a teaching assistant at ML4G camps twice, facilitated the AISES course from the Center for AI Safety, and co-founded EA Argentina.

  • Carlos Giudice

    Co-Director · Research Engineer, Equistamp

    Carlos is a Research Engineer at Equistamp and a co-director of BAISH. His move into a full-time AI safety role came directly out of an EA Global conference, the kind of path BAISH now helps its members take.

  • Luca De Leo

    Co-founding Director · Product Engineer, 80,000 Hours

    Luca is Product Engineer at 80,000 Hours and a co-founding director of BAISH. Involved in AI safety since 2016, he received grants from ACX+ and the Long Term Future Fund to upskill for AI safety research, worked in operations at Nonlinear, and won two first places and one second place at Apart Research research sprints. He previously ran operations for the AI Species YouTube channel, which has over 14 million views.

Want to talk to us?

Start with the contact form: tell us a bit about yourself and what you're looking for, and we'll point you to the right person or program.

Write to usOr book a call with Eitan↗

Leads

  • Nicolás Martorell

    Research Lead · Head of BAISH Labs

    Nicolás leads BAISH Labs. He is a postdoctoral researcher at UBA-CONICET's Applied Artificial Intelligence Laboratory, where he runs a research line on LLM interpretability, with peer-reviewed work at the xAI World Conference and an ICML 2026 workshop. He holds a PhD in Neuroscience from the University of Buenos Aires, spent four years as a teaching assistant and lead TA at Neuromatch Academy, facilitates BlueDot's Technical AI Safety Projects course, maintains the open-source concept-probe interpretability library, and is the author of ¿Qué es (y qué no es) la inteligencia artificial?, a general-audience book on AI published by Siglo XXI.

  • Nicolás SpinelliCommunications Lead
  • Marina RomanisioCommunity Lead
  • Rocío MonjesOperations Lead
  • Augusto Esquivel MasciottaCourses Lead · Frontier AI Governance

Team

  • Tobías BersiaTechnical AI Safety Project lead · Research Lead, BAISH Labs
  • Gonzalo HerediaTechnical AI Safety course owner · BAISH Labs
  • Gaspar LabastieCourse facilitator · ASTN platform · BAISH Labs
  • Tomás KorenblitCourse facilitator · BAISH Labs
  • Juan CharovskyCRM & automation
  • Santiago Nuñez RimedioInstitutional relations
  • Trinidad BorrellVolunteer
  • MB
    María BorrellAudiovisual content
  • DM
    Damián Barjau MateuWritten content
  • Josefina ZappettiniGraphic design
  • Aaron SznaiderBiosecurity community lead
  • Nahuel SegoviaCybersecurity community lead
  • TR
    Trinidad ReynosoFundraising
  • Malena CocozzellaVolunteer
  • Julián SzereszewskiResearch Fellow, BAISH Labs · Iliad alumnus
  • Lucio GarcíaResearch Fellow, BAISH Labs · PhD, University of Buenos Aires
  • Wendy BrauResearch Fellow, BAISH Labs
  • Tomás GiménezResearch Fellow, BAISH Labs

Advisors

  • Pablo StafforiniAI Enablement Manager, Epoch AI
  • Sergio AbriolaProfessor, University of Buenos Aires
  • Iván ArcuschinLead Research Scientist, Poseidon Research
  • Agustín Martínez SuñéPostdoc, University of Oxford · Formal verification × AI safety
  • Tobías MartinMIRI
  • Tao BurgaTechnology Fellow, Institute for Progress
  • Luis Enrique UrtubeyDirector, CEGIA
  • Guido BergmanFormer Courses Lead · Research Engineer, Equistamp
  • Mayra AmeneirosSenior Fellow, IBBIS · AI × biosecurity

Supported by

BAISH's work is made possible by:

Kairos

Pathfinder Program

Kairos supports AI safety field-building through its Pathfinder program, which funds BAISH's core operations.

Pathfinder↗Kairos↗

BlueDot Impact

Rapid grants

BlueDot funds our in-person course cohorts through per-cohort rapid grants, and awarded career transition grants to two of our members.

Visit website↗

grantmaking.ai

Launch Round

Funding for BAISH's operations through the grantmaking.ai launch round, with a follow-on grant from AISTOF via Manifund.

Visit website↗
Coefficient Giving

via AISAR

Coefficient Giving funds AISAR, our sister fellowship for technical AI safety research in Buenos Aires.

Visit website↗AISAR↗

Buenos Aires AI Safety Hub

Founded in 2024 · Buenos Aires

Join the WhatsApp community↗

Programs

  • Courses
  • AISAR fellowship
  • BAISH Labs
  • Resources

Organization

  • About
  • Research
  • Contact
  • Privacy Policy

Community

  • WhatsApp
  • Events calendar
  • Newsletter
  • Instagram
  • LinkedIn

Supported by: Kairos · BlueDot Impact · grantmaking.ai · Coefficient Giving

© 2026 BAISH. All rights reserved.