Database

No summary available yet.

Public Speaker, Author, Social Worker, Researcher, Systems Engineer

Zach Furman is a PhD student at the University of Melbourne working on singular learning theory and the mathematical foundations of deep learning, advised by Liam Hodgkinson and collaborating closely with Daniel Murfet and Timaeus. His research aims to make AI safer by understanding how neural networks work using tools from mathematics and physics, with a focus on developmental interpretability. He holds an undergraduate degree in mathematics and computer science from Boston University, and prior to his PhD he worked in rocket engineering (embedded software, electrical, and aerospace engineering) and briefly conducted machine learning interpretability and condensed matter physics research. He is affiliated with FAR.AI as a researcher, where he contributed to the "Eliciting Latent Predictions from Transformers with the Tuned Lens" paper. He also co-authored "The Loss Kernel: A Geometric Probe for Deep Learning Interpretability" and a position paper on singular learning theory for AI safety. He received a $40,000 grant from the Long-Term Future Fund in October 2023 to support six months of research in Daniel Murfet's group at the University of Melbourne, with results targeting publication at academic ML conferences.

4-month stipend to study refusals and jailbreaks in chat LLMs under Neel Nanda as part of the MATS 5.0 extension program

No summary available yet.

Philip Quirke (also known as PQ) is an AI safety researcher and Research Lead at Martian who made a notable career transition into AI safety from a background in software engineering, Agile coaching, and business analysis. He entered the field through Apart Research's hackathon and fellowship program, which he describes as a transformation, and subsequently served as a Research Project Manager at FAR.AI. His research focuses on mechanistic interpretability, including a widely-cited 2023 paper on how transformers perform integer addition and collaborative work on planning representations in recurrent neural networks trained to play Sokoban. He has co-authored papers on AI regulation and alternative AI architectures, and has produced five papers while securing approximately $90,000 in research grants. He received an LTFF grant to support a six-month career pivot into AI safety and alignment research, and has been involved with AI Safety Australia and New Zealand.

Mallory Strawn is Chief Operating Officer at TechCongress, overseeing the organization’s day‑to‑day operations. She is a seasoned professional with more than 12 years of experience in the tech industry focused on operations and knowledge management, holds a BBA from Georgia State University, and has supported numerous startups, SaaS companies, and nonprofit organizations.

No summary available yet.

6-month salary for Fabian Schimpf to upskill into AI alignment research and conduct independent research on limits of predictability

No summary available yet.

No summary available yet.

No summary available yet.

Amy Labenz serves on LASST’s board of directors and works as a director at the Centre for Effective Altruism, where she has previously served as general counsel. Before joining CEA she worked as a civil rights attorney in Detroit and as chief compliance officer and chief operating officer at the Machine Intelligence Research Institute, and she is a graduate of New York University School of Law.

Funding gap to pay for a video editor and scriptwriter

No summary available yet.

Brad Rogers Carson is an American lawyer and public servant who represented Oklahoma’s 2nd congressional district in the U.S. House of Representatives from 2001 to 2005 and later served as General Counsel and then Under Secretary of the Army, as well as Acting Under Secretary of Defense for Personnel and Readiness. He subsequently became the 21st president of the University of Tulsa and has since moved into leadership in AI policy advocacy and governance.

Bryce Meyer is a software engineer and the primary maintainer of TransformerLens, the leading open-source library for mechanistic interpretability research on GPT-style language models. TransformerLens was originally created by Neel Nanda and allows researchers to load 50+ open-source language models and inspect their internal activations, making it the de facto standard tool for mechanistic interpretability work at organizations including Anthropic, Meta Research, Redwood Research, and Apollo Research. Meyer has maintained the library with a track record of consistent contributions, rapid iteration to support newly released models, and active community support via a weekly live-coding stream in the Open Source Mechanistic Interpretability Slack. He received a $50,000 grant from the Long-Term Future Fund in 2023 to build and enhance open-source mechanistic interpretability tooling, followed by a $90,000 year-long LTFF stipend to serve as TransformerLens's primary maintainer. He is also the president of Pomelo Productions, an independent software development studio based in Milwaukee, Wisconsin, and a self-taught developer with many years of professional engineering experience.

6-month stipend to continue independent AI alignment research from MATS 5.0 on situational awareness and deception

No summary available yet.

Independent AI safety researcher investigating structural failure modes in agentic systems, including logic-layer escapes and governance enforcement. Developer of EasyStreet / AEGIS-ALD-W1, a deterministic, audit-grade evaluation framework for AI agents.

Co-founder and CEO of Noya, a direct air capture company aiming to reverse climate change by pulling CO2 from the atmosphere; he studied Chemical‑Biological Engineering at MIT and previously worked as a project manager at Tesla and Harley‑Davidson.

AI alignment researcher who has collaborated with Orthogonal on the QACI formal-goal alignment agenda, co-authoring the “formalizing the QACI alignment formal-goal” article.

No summary available yet.

JJ Hepburn is the founder of Ashgro and previously worked as a facilitator at AI Safety Camp; he has training in machine learning with TensorFlow on Google Cloud Platform and studied at Macquarie University and the Australian National University.

No summary available yet.

No summary available yet.

No summary available yet.

No summary available yet.

No summary available yet.

AI RnD Lead and Research Supervisor, Writer, Composer

No summary available yet.

No summary available yet.

Funding for (academic/technical) AI safety community events in London

No summary available yet.

No summary available yet.

No summary available yet.

Johannes C. Mayer is an independent AI alignment researcher and game developer who has dedicated his career to addressing AI existential risk. He participates in the AI safety research community through LessWrong and the AI Alignment Forum, where he has published over 76 posts on topics including structural approaches to alignment, computational models of intelligence, and world model interpretability. He completed the MATS Summer 2022 cohort under the mentorship of Evan Hubinger, and has served as a mentor for the Supervised Program for Alignment Research (SPAR) at UC Berkeley. His research agenda focuses on translating intuitive concepts such as goals, wanting, and abilities into formal computational frameworks, and on constraining AI reasoning processes structurally rather than purely specifying outcome-level objectives. He received a grant from the Long-Term Future Fund to pursue this research on turning intuitions about intelligence into concepts applicable to computational systems.

One year of bootstrapped development, four patent filings, seeking support to continue.

No summary available yet.

Tim Farrelly is a founder and engineer based in Dublin, Ireland. He studied Computer Engineering at Trinity College Dublin and previously conducted AI safety research with academics at Oxford and Trinity College Dublin, focusing on cooperative AI and multi-agent systems. He co-founded Field of Vision, a company that creates haptic devices to help visually impaired sports fans experience live games, which was featured in Time Magazine's Best Inventions. He also co-founded Patch, an initiative supporting talented Irish youth, and Hack Ireland/Hack Europe, Ireland's largest student hackathon. He has received a grant from the Long-Term Future Fund for equipment to improve productivity while pursuing AI safety research, and his personal website states he is refocusing his energy on AI safety work.

Aric Floyd is the on‑camera host of AI In Context, the YouTube channel produced by 80,000 Hours that uses long‑form documentary‑style videos to explain transformative AI and its risks. He joined 80,000 Hours in 2025 as an Associate Video Producer after previously working as a film and television actor based in Los Angeles while on extended leave from undergraduate physics studies at Stanford University.

Peter Mühlbacher is a founding research scientist at FutureSearch. He holds a PhD in Mathematics from the University of Warwick and previously worked as a research scientist at Metaculus and as a risk analyst at UniCredit Bank Austria.

No summary available yet.

Strategy advisor to EA Hungary and AI Safety Hungary, supporting organisational strategy and operations for their EA and AI safety field-building projects.

No summary available yet.

Making AI safety & security scalable

Events and partnerships professional working on AI governance initiatives; Events and Partnerships Lead at Intelligence Rising, also a facilitator at Technology Strategy Roleplay, and previously Chief of Staff to a senior AI policy practitioner with appointments at OECD.AI, PECC (France), UNESCO and the Global Partnership on AI.

Anastasiia Gaidashenko is a Project Manager in the Office of the CEO at FAR.AI and previously worked as a Technical Program Manager and Data Scientist on responsible AI governance and applied machine learning.