P.H.I
P.H.I. (Prompt Human Inc.) was a French sole-proprietorship research entity operated by Quentin Feuillade-Montixi, a SERI MATS 3 alumnus and former METR contractor. The entity received Long-Term Future Fund (LTFF) grant funding in late 2023 to support six months of research on model psychology and evaluation, with outputs published to LessWrong. Feuillade-Montixi developed the LLM Ethology research agenda, studying large language models through a behavioral and cognitive science lens, with applications to red-teaming and AI safety evaluation. The entity was dissolved in December 2023 as Feuillade-Montixi moved on to co-found PRISM Eval.
Funding Details
- Annual Budget
- -
- Monthly Burn Rate
- -
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- $43,159
- Fiscal Sponsor
- -
Theory of Change
P.H.I.'s research operated on the premise that understanding LLM cognition and behavior in depth — through a psychology and ethology lens — is a prerequisite for building robust AI safety evaluations. By characterizing how and why LLMs behave in different contexts, researchers can better predict failure modes, design targeted evaluations, and develop alignment interventions. Published findings on LessWrong were intended to build shared understanding in the AI safety research community and inform downstream red-teaming and evaluation work.
Grants Received
from Long-Term Future Fund
Projects
No linked projects.
People
No linked people.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.
Details
- Last Updated
- Mar 21, 2026, 10:50 PM UTC
- Created
- Mar 19, 2026, 10:42 PM UTC