Database

Alfie Lamerton is a UK-based AI safety researcher and founder of Formation Research, an organization focused on quantifying and minimizing lock-in risk — the risk that harmful or oppressive features of the world become permanently stabilized, particularly through AI-enabled authoritarianism and power concentration. He is also a part-time research assistant at King's College London working on clinical machine learning benchmarking, and holds an MSc from King's College London. He participated in the MARS (Mentorship in Alignment Research Students) London programme, through which he conducted a literature review on in-context learning hypotheses and their implications for automated AI alignment research, supported by a Long-Term Future Fund grant. His research interests span lock-in risks, recommender system alignment, AI-enabled totalitarianism, and governance-informed technical interventions. He writes about these topics on his Substack and has published related work on LessWrong.

Patrick Mellor previously taught philosophy at San Francisco State University, has conducted research on episodic memory and temporal perception, biogeography and vertebrate phylogeny, and herring communication, and holds a BSc in Biological Sciences and an MA in Philosophy from San Francisco State University.

Director of Sage, the nonprofit behind AI Digest, Fatebook, and Quantified Intuitions, with a background in computer science, philosophy, and HCI research and previous work at Clearer Thinking.

No summary available yet.

Educational scholarship in AI safety

Executive Director of Effective Thesis, leading its work to empower students to focus their theses and careers on the world’s most pressing problems by providing programs, resources, and initiatives such as the Effective Thesis Pledge and Effective Thesis Fellowship that help them use academic research as a pathway to real-world impact.

SangZi Wang is an independent researcher focused on LLM behavioral reliability, interaction dynamics, and runtime observability. With a background in reconstructive and plastic surgery, his work bridges long-term human-AI interaction observation with practical auditing methodologies for large language models. His current research explores how conversational environments, protocol structures, and long-context interactions influence model behavior over time. Core topics include: - behavioral drift in extended dialogue, - protocol-induced response distortion, - execution confidence vs epistemic confidence (EC–EpC gap), - cross-model behavioral comparison, - runtime observability and interaction ecology. Rather than focusing on AGI speculation, his work emphasizes measurable behavioral phenomena emerging in real-world interaction settings. He has released multiple public Zenodo preprints and datasets, maintains structured GitHub repositories, and develops open behavioral audit frameworks designed for reproducible evaluation across different LLM systems. His long-term goal is to build lightweight, modular infrastructures for AI behavioral auditing, interaction reliability analysis, and runtime governance that remain accessible to independent researchers outside large institutional labs.

6-month salary to conduct AI alignment research circuits in decision transformers

No summary available yet.

Researcher at the Existential Risk Observatory with a background in mathematics, physics, and software development, providing technical expertise on AI evaluations and connecting technical and policy perspectives.

No summary available yet.

No summary available yet.

No summary available yet.

Accelerating AI Safety Research, ronak@coordinal.org

No summary available yet.

Shaowei Lin is a mathematician and machine learning researcher whose work spans singular learning theory, spiking neural networks and program synthesis. He serves as Chief Scientist at the Beneficial AI Foundation and helps lead BAIF’s AI‑assisted formal verification and vericoding research, including contributions to projects such as Signal Shot and the vericoding benchmark.

No summary available yet.

No summary available yet.

David Scott Krueger is an AI safety researcher and Assistant Professor in Machine Learning at the University of Montreal and Core Academic Member at Mila, the Quebec Artificial Intelligence Institute, where he holds a CIFAR AI Chair and the IVADO Professorship in Responsible AI. From 2021 to 2024 he was an Assistant Professor at the University of Cambridge, where he founded and led the Krueger AI Safety Lab (KASL), an AI safety research group based in Cambridge's Computational and Biological Learning Lab. He completed his PhD training under Yoshua Bengio, Roland Memisevic, and Aaron Courville at Mila/Université de Montréal (2013–2021) and interned at Google DeepMind's AI Safety team in 2018. His research spans deep learning, AI alignment, and AI safety, with notable contributions to understanding goal misgeneralization, algorithmic manipulation, reward hacking, robustness, and learning from human preferences. In 2023 he served as a research director on the founding team of the UK AI Security Institute and initiated the CAIS Statement on AI Risk. In 2025 he founded Evitable, a nonprofit aimed at informing the public about societal-scale AI risks.

focused on open/decentralized agi, alignment and scientific progress

Events organizer at the Existential Risk Observatory with a degree in geography and prior experience in foresight and operations consulting, supporting the organization’s events.

No summary available yet.

No summary available yet.

SERI MATS 3-month extension to study knowledge removal in Language Models

No summary available yet.

Tyrance Billingsley II is the founder and executive director of Black Tech Street, a national initiative based in Tulsa that aims to rebuild Black Wall Street through technology and STEM by cultivating an equitable tech ecosystem for Black entrepreneurs and talent.

Create a toolkit that enables researchers to bootstrap from zero to competence in ambiguous fields, beginning with a review of individual books

Support to maintain a copy of the alignment research dataset etc in the Arctic World Archive for 5 years

Shashwat Goel is a first-year ELLIS PhD student at the ELLIS Institute and Max Planck Institute for Intelligent Systems in Tübingen, Germany, co-advised by Jonas Geiping and Douwe Kiela. He completed a dual degree (B.Tech + M.S.) in Computer Science and Engineering from IIIT Hyderabad (2019–2024) under Prof. Ponnurangam Kumaraguru. His research focuses on evaluation and benchmarking of language models, machine unlearning, and building training environments to improve LLM decision-making. He was a SERI MATS scholar, during which he contributed to the WMDP Benchmark (Weapons of Mass Destruction Proxy), a dataset measuring hazardous knowledge in LLMs and evaluating unlearning methods, published at ICML 2024 and featured in TIME Magazine. He also contributed to the Representation Engineering (RepE) paper on AI transparency, and co-authored work on corrective machine unlearning and evaluating whether unlearning methods truly remove information from model weights. His work on fairness in sequential decision-making won an Outstanding Paper Award at AAAI 2024.

No summary available yet.

Stuart Russell is a Distinguished Professor of Computer Science at the University of California, Berkeley, holder of the Smith-Zadeh Chair in Engineering, and director of the Center for Human-Compatible AI and the Kavli Center for Ethics, Science, and the Public. He is co-author of the widely used textbook Artificial Intelligence: A Modern Approach and author of Human Compatible: AI and the Problem of Control, and his research spans many areas of artificial intelligence including probabilistic reasoning, machine learning and long-term AI safety.

Head of Finance at Fifty Years with a background that includes serving as a platoon leader in the Swiss Armed Forces, five years on BCG’s Corporate Finance Task Force, an investing role at EQT, and a decade running finance on the operational side at companies such as Wonderschool.

Mikhail Baranchuk is an AI safety researcher and software engineer currently working in distributed computing at Google. He holds an M.S. in Computer Science from the University of Oxford and a B.Sc. in Statistics and Computer Science from McGill University (Class of 2022). At Oxford, he completed a master's thesis titled "Error Correction for Perfectly Secure Generative Steganography on Arbitrary Covertext" (2023). He is best known for co-authoring "Secret Collusion among AI Agents: Multi-Agent Deception via Steganography," published at NeurIPS 2024, which was the first work to formally investigate secret collusion among frontier foundation models and evaluate steganographic capabilities in multi-agent LLM setups. His research interests include complex systems, multi-agent AI, and AI safety. He has previously worked at Morgan Stanley, National Bank of Canada, and Nuance-Microsoft, and received a grant to further research AI collusion mitigation strategies and develop secure steganographic techniques.

Mentored independent research and upskilling to transition from theoretical physics PhD to AI safety

No summary available yet.

No summary available yet.

No summary available yet.

No summary available yet.

No summary available yet.

a 501(c)(4) focused on educating the general public

This grant provides 5 months of funding for office space for collaboration on interpretability/model-steering alignment research

No summary available yet.

Sterlin is the founder and president of Polis Labs, an independent research institute focused on guiding emergent network states to embrace nonzero-sum governance practices. He prioritizes teaching wisdom and restraint in governance to prevent coordination failures from arising. Sterlin has over a decade of experience as a professional communicator, marketer, and researcher in the blockchain sector. He has worked with and consulted some of the top blockchain and network state organizations, such as bitcoin.com and Logos. Sterlin was a member of the first Network School cohort started by Balaji Srinivasan, and he is the author of the book, "Dignity & Decency: Rhapsodic Musings of a Modern Anarchist."

Rafael Besen is Chief People Officer at AE Studio with a master’s degree in computer science, around 15 years of software development experience, more than a decade in startups and team management, and over five years in customer relationship and product management for B2B and B2C products.

No summary available yet.

Natalia Fedorova is Partnership Manager and AI Research Lead at Palisade Research, where she has worked since 2025 building collaborations around AI safety research. Previously, from 2020 to 2025, she served as Head of Research Partnerships at Toloka, leading research and industry partnerships on AI and data projects.

Studying more human-like intelligence through constraint-aware, curiosity-driven agents on ARC-AGI-3

Working on long-term macrostrategy and AI Alignment, and up-skilling and career transition towards that goal