Iván Arcuschin Moreno
Bio
Iván Arcuschin Moreno is an AI safety researcher based in London, UK, currently serving as Lead Research Scientist at Poseidon Research, a US-based AI safety non-profit where he heads the London research team. He holds a Computer Science PhD from the University of Buenos Aires, Argentina (2018–2024), where his thesis focused on automated test generation for Android apps. He completed two terms at the ML Alignment & Theory Scholars (MATS) program: the first (Jan–Jul 2024) under Adrià Garriga-Alonso at FAR AI, producing InterpBench, a collection of 86 semi-synthetic transformers with known circuits for evaluating mechanistic interpretability techniques (NeurIPS 2024); and the second (Jan 2025–Feb 2026) under Arthur Conmy at Google DeepMind, focusing on chain-of-thought faithfulness. He is lead author on "Chain-of-Thought Reasoning In The Wild Is Not Always Faithful" (ICLR 2025 workshop, 140+ citations within a year) and a contributor to the MIB mechanistic interpretability benchmark (ICML 2025). He also co-authored a position paper with Yoshua Bengio and founded AI Safety Argentina (AISAR), a research scholarship program to grow the AI safety research community in Latin America, supported by a $77,000 grant from Coefficient Giving.
Links
- Personal Website
- https://iarcuschin.com/
- Twitter / X
- LessWrong
- -
Grants
from Long-Term Future Fund
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.
Details
- Last Updated
- Mar 22, 2026, 4:30 PM UTC
- Created
- Mar 20, 2026, 2:51 AM UTC