Samuel Nellessen
I want to help make AI go well. I work on scalable monitoring of LLM agents.
I’m joining Refractal in London as Lead Research Engineer. Previously, I worked on a paper about evaluation awareness and detecting alignment faking at LASR Labs, mentored by Felix Hofstätter (Apollo Research).
Recent Updates
- Joining Refractal in London as Lead Research Engineer, working on scalable agentic monitoring.
- Incoming LASR Labs Fellow in London.
- Presented Tag-Along Attacks at Foresight Vision Weekend UK.
- Started working from the Foresight Node Berlin as an independent AI Safety grantee.
- Awarded a $22.5k Foresight Institute AI Safety Grant.
- Released David vs. Goliath on arXiv.
- Punctuation and Predicates in Language Models accepted to Findings of EACL 2026.
- Joined KachmanLab to work on automated red-teaming.
- Completed ARENA 5.0 in London.
Selected Papers
David vs. Goliath: Verifiable Agent-to-Agent Jailbreaking via Reinforcement Learning
Punctuation and Predicates in Language Models
Controllability, not predictability, flexibly gates the impact of Pavlovian bias on action selection
Selected Talks
Tag-Along Attacks: evaluating what AI agents do, not just what they say
Foresight Vision Weekend UK 2026 - VIP Gathering
One-minute lightning talk on verifiable red-teaming for tool-using LLM agents.
Tag-Along Attacks: evaluating what AI agents do, not just what they say
Foresight Vision Weekend UK 2026
Conference lightning talk on behavioral evaluation beyond textual refusals.
Tag-Along Attacks: For LLMs, by LLMs, with LLMs
Foresight Institute AI Salon Berlin
Talk on Tag-Along Attacks and LLM-driven red-teaming for tool-using LLM agents.
Projects and Open Source
Styx Interchange: Localizing Refusal Behavior in LLMs
Mechanistic interpretability sprint comparing activation patching against input gradients to localize causal refusal behavior across layers.
UK AISI inspect_ai contribution
Contributed a vLLM startup behavior fix to improve reliability in AI evaluation infrastructure.
Prime Intellect Environments
Built a text-retrieval environment and benchmark utilities for agentic AI benchmarking.
ARENA 5.0 Capstone: Internal Representations in SONAR Autoencoders
Research sprint probing whether SONAR text autoencoders encode correctness across code, grammar, arithmetic, and chess domains.
Selected Writing
Investigating Internal Representations of Correctness in SONAR Text Autoencoders
A capstone writeup on probing whether compressed multilingual representations carry signals for code validity, grammaticality, arithmetic, and chess syntax.
Brain enthusiasts in AI Safety
A guide for students of cognitive science and neuroscience considering work in AI safety.