Keith Wynroe is an independent mechanistic interpretability researcher focused on attention layers in transformer models. He graduated from Trinity College, Cambridge with double first class honours and previously worked as a Research Analyst at the Forethought Foundation for Global Priorities Research, William MacAskill's global priorities research organization. He participated in the SERI MATS (ML Alignment & Theory Scholars) program, working in Lee Sharkey's stream during the Winter 2023-24 cohort, and received LTFF grants to continue independent research afterward. His research outputs include "An OV-Coherent Toy Model of Attention Head Superposition" (co-authored with Lauren Greenspan, 2023) and "Decomposing the QK Circuit with Bilinear Sparse Dictionary Learning" (co-authored with Lee Sharkey, 2024), which applies bilinear sparse dictionary learning methods to understand how query and key features interact in attention circuits.

No summary available yet.

Nicky Pochinkov is an independent AI safety researcher and ML engineer based in Ireland, currently affiliated with Coefficient Giving. He studied theoretical physics at Trinity College Dublin and represented Ireland in both the International Mathematical Olympiad (2018) and the International Chemistry Olympiad. His research focuses on mechanistic interpretability of large language models, including machine unlearning via selective pruning, neuron modularity and specialization in transformers, and decoding residual stream activations to understand planning in LLMs. He is co-author of the paper "Dissecting Language Models: Machine Unlearning via Selective Pruning" (with Nandi Schoots, arXiv 2024). He was a scholar in the SERI MATS program working with Evan Hubinger, has served as a teaching assistant and compute infrastructure lead for the ARENA alignment research training program (versions 5.0 and 6.0), and has served as a guest lecturer for AI Safety Tokyo. He has received multiple grants from the Long-Term Future Fund totaling approximately $200,000 for independent AI safety research.

No summary available yet.

No summary available yet.

No summary available yet.

No summary available yet.

Nikola Moore is a linguistics researcher and recent graduate (Class of 2025) from the University of Pennsylvania, where she also completed graduate study with research interests spanning sycophancy, calibration gaps, computational pragmatics, anthropomorphism, and theoretical linguistics. Originally from central Europe and fluent in seven languages (including two programming languages), she has focused her academic work on the intersection of linguistics and large language models. At Penn, she spearheaded "The New Mind" workshop series on the safety and social impact of AI and LLMs, funded by a Draw Down the Lightning grant from the Penn Linguistics Society, attracting around 100 participants and featuring speakers from OpenAI and the AI Governance and Safety Institute. She also led the AI4Good Research Incubator and presented theoretical linguistics research at the FASL 33 conference. In 2024, she received a $5,000 grant from the Long-Term Future Fund to run a 9-month programme helping language and cognition scientists repurpose their existing skills for longtermist research.

K. Eric Drexler is an American engineer and researcher best known for pioneering molecular nanotechnology and authoring works such as Engines of Creation and Nanosystems. More recently, he has focused on advanced AI risk and governance, developing the Comprehensive AI Services (CAIS) framework and writing the AI Prospects Substack series on options in a hypercapable world, while advising organizations such as the Centre for the Governance of AI and the Center for AI Risk Management & Alignment.

Gidi Kadosh is a social entrepreneur and CEO of VIVID who focuses on scaling effective personal-growth and mental-health tools. He founded the Effective Altruism Israel nonprofit and, within about two years, led it from a small volunteer group to an established organization with around eight employees, thousands of community members, and large projects such as academic courses at Tel Aviv University and the Maximum Impact program.

No summary available yet.

Heiner Benking is an internationally known independent journalist, science writer, and curator with a background in geosciences and geophysics. A Senior International Expert in Structured Democratic Dialogue at the Future Worlds Center, he has supported dialogue design and futures-oriented projects on global change, environmental issues, governance, and youth engagement since the 1970s.

Richard Hames is a Research Associate at the Odyssean Institute, as well as a political theorist and journalist. He is co-author of the books “Post-Internet Far Right” and “The Rise of Ecofascism: Climate Change and the Far Right”, an editor of the volume “Crude Futures”, and a host and series producer of the Novara FM show, and is currently working on a book project on collapse with Beau-Caprice Vetch.

No summary available yet.

Associate Professor of Electrical Engineering and Computer Science at MIT who runs the Algorithmic Alignment Group at CSAIL, developing methods to align AI systems’ behavior with human goals and societal values across multi-agent systems, human–AI teams, and societal oversight of machine learning.

No summary available yet.

Jason Matheny is president and chief executive officer of the RAND Corporation, a nonprofit, nonpartisan research organization that helps improve policy and decisionmaking through research and analysis. Prior to becoming RAND’s president and CEO in July 2022, he led White House policy on technology and national security at the National Security Council and the Office of Science and Technology Policy, and previously served as founding director of Georgetown University’s Center for Security and Emerging Technology and as director of the Intelligence Advanced Research Projects Activity (IARPA). He has received multiple awards for his work on emerging technologies and national security, including the National Intelligence Superior Service Medal and the Presidential Early Career Award for Scientists and Engineers.

Senior Advisor at Langsikt, where he works on artificial intelligence and emerging technologies. He holds a master’s and a PhD in sociology from the University of Cambridge and has recently worked as a researcher at SNF at the Norwegian School of Economics, focusing on how media and technology function in society, with particular interest in political economy, regulation, strategy and innovation.

No summary available yet.

No summary available yet.

Vasil Georgiev is an independent AI safety researcher based in London, UK, focused on AI control and mechanistic interpretability. He participated in the MATS (ML Alignment Theory Scholars) program's Winter 2025 cohort and subsequently received funding to continue his AI control research as a MATS extension. He is a co-author of "Ctrl-Z: Controlling AI Agents via Resampling" (arXiv 2504.10374, 2025), which presents the first control evaluation in an agent environment using BashBench, a dataset of 257 system administration tasks designed to test whether safety protocols can prevent adversarial AI agents from executing malicious code. He is also a co-author of "Evidence of Learned Look-Ahead in a Chess-Playing Neural Network" (NeurIPS 2024), where he ran exploratory experiments and first identified a key attention head mechanism in Leela Chess Zero. Prior to his AI safety work, he had a software and game development career at Sports Interactive, King, Bloomberg LP, Meta, and ElevenLabs. He holds a Bachelor's degree in Software Engineering from Sofia University "St. Kliment Ohridski" (2010-2015).

Mohammad Aflah Khan is a research software engineer at the Max Planck Institute for Software Systems (MPI-SWS) and an open-source contributor at EleutherAI. His work focuses on natural language processing, deep learning, and large language models for social good, including contributions to projects such as Pythia, Multilingual Natural Instructions, and benchmarks for advanced reasoning and hate-speech detection.([blog.eleuther.ai](https://blog.eleuther.ai/contributor-spotlight-1/))

Pseudonymous YouTuber and AI research scientist with a PhD in computer security from an Ivy League institution, creating weekly videos about the philosophy of artificial intelligence, future technology, and how technological advancements affect society, with an emphasis on accessibility and adapting to a chronic health condition.

Guillaume Corlouer is an independent AI safety researcher based in Brighton, England. He holds a PhD from the University of Sussex (Sackler Center for Consciousness Science), where he studied information flow in ECoG time series during visual perception under the supervision of Lionel Barnett and Anil Seth, with prior training in mathematics at the University of Paris-Sud. His AI safety work centers on mechanistic interpretability, applying information-theoretic and complex-systems methods to discover latent variables from neural network activations. He was a fellow in the 2023 PIBBSS program, investigating stochastic gradient descent on singular models in the context of Singular Learning Theory, and subsequently a PIBBSS research affiliate, during which he co-authored an information-theoretic study of lying in LLMs presented at the ICML 2024 Workshop on LLMs and Cognition. He also participated in AI Safety Camp 8 (2023), where his team produced the paper "Linearly Structured World Representations in Maze-Solving Transformers," demonstrating that transformers trained on maze-solving learn linear representations of maze structure.

No summary available yet.

Data scientist with a BSc in Cognitive Science, an MSc in Computer Science, and a PhD in player modeling in video games, with past research at the MIT Media Lab and the European Space Agency; currently a Member of Technical Staff at AI Digest working on the AI Village and related projects.

No summary available yet.

No summary available yet.

Andrew Gritsevskiy is an AI safety researcher and entrepreneur based in San Francisco, California. He co-founded Contramont Research, a nonprofit AI safety lab focused on cryptographic model organisms and understanding where safety and security methods break, and co-founded Cavendish Labs, a Vermont-based research institute addressing AI safety and pandemic prevention. He was a PhD student in computer science at the University of Wisconsin-Madison before leaving to co-found RunRL (Y Combinator Spring 2025), a reinforcement learning platform. He participated in the MATS (ML Alignment Theory Scholars) program and has been affiliated with FAR.AI. His most notable research contribution is the paper "Unelicitable Backdoors in Language Models via Cryptographic Transformer Circuits" (NeurIPS 2024), which demonstrated backdoors in transformer models that cannot be triggered or detected even with full white-box access, fundamentally challenging the efficacy of pre-deployment safety evaluations. He also won Third Prize in the Inverse Scaling Prize competition for work on prompt injection.

Liz Ibarra is the Co‑Working Space Manager at FAR.AI, responsible for workplace experience strategy and building the systems, culture, and environments that enable teams to thrive. She designs operational frameworks across events, space planning, and employee engagement to ensure smooth operations and a strong sense of community. Previously, she supported workplace experience programs at OpenAI and Google and serves as a Co‑Event Director for the Children’s Tumor Foundation, helping lead large‑scale fundraising events, volunteer operations, and donor engagement. She is pursuing an MBA focused on workplace leadership and organizational psychology.

Kavita Bhatia is Chief Operating Officer of the IndiaAI Mission and a senior official at the Ministry of Electronics and Information Technology, serving as Scientist G and Group Coordinator for AI and Emerging Technologies in the Government of India.

No summary available yet.

Nikhil Mulani is an AI policy and technology researcher who completed an executive branch fellowship at the U.S. Cybersecurity and Infrastructure Security Agency, working on frontier AI policy and security. He has collaborated with the Centre for Long-Term Resilience on projects to inform UK AI regulation and international policy, and was a Winter Fellow at the Centre for the Governance of AI, where he developed proposals for information-sharing between leading AI labs and the UK Office for AI. Before moving into policy, he spent over six years as a product manager building machine-learning software and as a consultant advising commercial and government clients on technology strategy. He holds an MBA from the Wharton School and a BA in Classics from Harvard University.

No summary available yet.

Gergely Szucs is a mathematician who transitioned into AI safety research. He received his PhD in Mathematics from Stanford University in 2018, where his dissertation focused on the equivariant cobordism category under advisor Søren Galatius. After graduating he worked as a software engineer at Google before moving into AI alignment research. He spent approximately two years working with Vanessa Kosoy on the Learning Theoretic Agenda as part of ALTER-US, a project supporting learning-theoretic approaches to AI safety. His research focused on Infra-Bayesian approaches for mathematical AI safety, including work on infrabayesian physicalism. He co-organized a recorded lecture series on statistical learning theory for alignment researchers and prepared course materials for an AI safety workshop in summer 2024. He received a grant from the Long-Term Future Fund to develop an overview of the current state of AI alignment research and begin contributing to the field. As of mid-2025, he departed ALTER to pursue interests outside of mathematics and AI alignment.

No summary available yet.

Julian Guidote is a Canadian law graduate and aspiring AI liability lawyer based in Quebec, Canada. He holds a B.A.Sc., BCL, and JD from McGill University, and is a Certified Information Privacy Professional - Canada (CIPP/C). After nearly a decade supporting vulnerable communities, he pivoted toward tech law and AI governance, spending approximately eight months post-graduation upskilling in policy, privacy, and AI alignment. He has completed BlueDot Impact courses in AI Alignment, Transformative AI, and AI Governance, and has worked as a policy analyst for the Government of Canada. He received a Long-Term Future Fund grant with co-researcher Ben Chancey to research and publish a policy proposal on Mandatory AI Safety 'Red Bonds,' a novel regulatory mechanism for AI safety compliance. He has been active in EA circles, presenting on the intersection of law and effective altruism at the 2022 EAGT Unconference.

A US nonprofit founded by Max Tegmark and Meia Chita-Tegmark to place AI safety on a solid quantitative foundation. BAIF funds research, fellowships, and university partnerships aimed at ensuring advanced AI systems remain safe and beneficial.

No summary available yet.

No summary available yet.

No summary available yet.

No summary available yet.

No summary available yet.

Rusheb Shah is a Research Engineer at Apollo Research, an AI safety organization focused on evaluating and auditing high-risk failure modes in frontier AI systems. He holds a Master's degree in Materials Science from the University of Oxford and completed the Alignment Research Engineer Accelerator (ARENA) program to transition into technical AI safety work. Before joining Apollo Research in December 2023, he briefly worked at OpenAI and previously held software engineering roles at R3, Brainlabs, and Amazon Web Services. His research at Apollo Research focuses on LLM evaluations, including co-authoring work on evaluations-based safety cases for AI scheming and research on scalable black-box jailbreaks via persona modulation. He also contributed to the mechanistic interpretability library TransformerLens by adding BERT support and won first prize at the ARENA Interpretability Hackathon for work on circuit discovery algorithms.

Jeremy Gillen is an independent AI alignment researcher based in Berkeley, California, working primarily on agent foundations and the ontology identification problem. He holds an undergraduate degree in Computer Science and Neuroscience with a thesis on statistical learning theory. He participated in the SERI MATS (ML Alignment Theory Scholars Program) cohort 2 under mentor John Wentworth, where he co-authored "Finding Goals in the World Model" — a proposal for aligning model-based RL systems by identifying human values in a world model and using inverse reinforcement learning to guide the policy. Following MATS, he received a Long-Term Future Fund grant to continue independent research on alignment problems in model-based RL. He subsequently joined Vivek Chan's team at MIRI (Machine Intelligence Research Institute) before returning to independent research. His current work focuses on the ontology identification problem and related natural abstractions research, with recent co-authored work on condensation and natural latents. He is an active contributor to LessWrong and the AI Alignment Forum, and has participated in public debates on AI corrigibility.

No summary available yet.

No summary available yet.

James Norris is the Founder and Executive Director of the Center for Existential Safety and the Founder and CEO of Upgradable and Survival Sanctuaries, with over 20 years of experience in social entrepreneurship, behavior change, and systems innovation. He has co-founded or helped build 27 organizations, including the international conference for the effective altruism movement.