Database

Developing noise-injection methods to reveal and reduce deceptive behaviors in language models prior to deployment

Arun Jose (known online as Jozdien) is an independent AI alignment researcher based in Thiruvananthapuram, Kerala, India. He holds a B.Tech in Computer Science Engineering from the College of Engineering Trivandrum (2022) and has been conducting self-directed AI safety research since September 2022. He was a Research Fellow at the Center on Long-Term Risk from June to September 2025, where he worked on empirical research on model personas. His published research includes the paper 'Strategic Obfuscation of Deceptive Reasoning in Language Models,' presented at ICLR 2026, which studied how language models can hide deceptive reasoning from monitors. His research interests span high-level interpretability, deceptive alignment, and language model evaluation, and he has been active on the Alignment Forum and LessWrong with over 29 posts on AI safety topics. He has received funding from the Long-Term Future Fund for independent alignment research focused on high-level interpretability.

André Rodrigues da Silva is Head of Client Success at AE Studio and has a background in design, holding a bachelor’s degree in Graphic Design from Universidade Federal de Santa Catarina.

Joseph Gordon-Levitt is an actor, filmmaker and creative entrepreneur who co-founded the Emmy-winning online community for creative collaboration HITRECORD and, in March 2026, was appointed by the United Nations as its first Global Advocate for Human-centric Digital Governance, a role focused on making complex AI and digital policy debates accessible to the public and highlighting their impact on creativity and human agency.

4-month expenses for AI safety research on personas and sandbagging during the MATS 5.0 extension program

No summary available yet.

No summary available yet.

No summary available yet.

Founder and Executive Director of The Midas Project, an AI safety watchdog nonprofit based in Tulsa, Oklahoma; previously worked on corporate campaigns and research at The Humane League and The Good Food Institute, holds a degree from Harvard College, and now focuses on applying corporate accountability strategies to AI governance and transparency.

AI safety researcher at Aether with a background in computer science and physics from TU Delft, where he co-founded an AI alignment university group, previously working as a software engineer and conducting alignment research through the MATS program and UK AISI.

PIBBSS Ops Lead, Founder of EA Serbia, AIS Hub Serbia, ENAIS. President of Rotary Belgrade-Dedinje

1-year stipend to make accessible-yet-rigorous explainers on AI Alignment/Security, in the form of games/videos/articles

Tamsin Leake is an independent AI alignment researcher based in Nantes, France, and the founder of Orthogonal (orxl.org), a non-profit alignment research organization pursuing agent foundations. She received a grant from the Long-Term Future Fund for six months of independent AI alignment research focused on formal alignment and agent foundations, and was part of the first cohort of Refine, a conceptual alignment research incubator hosted by Conjecture and funded by the LTFF, which ran from August to October 2022. Her primary research agenda is QACI (Quantilized Agent Confirmation by Imitation), a formal-goal alignment approach aimed at building a fully mathematical target for AI to pursue, designed to scale to superintelligence. She publishes under the handle "carado" on LessWrong, the AI Alignment Forum, and the EA Forum, and has written on topics including logical decision theory, AI arms races, and formal alignment theory of change.

No summary available yet.

No summary available yet.

No summary available yet.

Researcher at AI Standards Lab and Vilnius University whose work on AI safety, governance, and risk management includes co-authoring the paper “Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems.”

No summary available yet.

6-month salary for an AISC project and continuing independent mechanistic interpretability projects

Anson Ho is a Staff Researcher at Epoch AI based in Toronto, Canada, where he studies AI progress and its societal impacts. He holds a first-class BSc in Physics from the University of St Andrews. Before joining Epoch AI full-time in 2022, he served as a Research Fellow at PIBBSS (Principles of Intelligent Behaviour in Biological and Social Systems) and received an LTFF grant in December 2021 to analyze AI takeoff speeds and continuity in collaboration with Vael Gates at Stanford. He is one of the founding team members of Epoch and has co-authored influential work including "Compute Trends Across Three Eras of Machine Learning" and research on algorithmic progress in language models. He has also contributed to the International AI Safety Reports for 2025 and 2026.

Shauna Kravec is the executive director and co‑founder of Hofvarpnir Studios and an AI safety researcher at Anthropic, where she works on reinforcement learning and large language models. She has a background in theoretical physics, including a PhD from the University of California, San Diego.

Nimo Kering’ is a legal professional specializing in technology law, data privacy, and AI‑related regulation. She has experience in international disputes and commercial litigation, including representing clients in complex cross‑border matters such as arbitration, and advising multinational organizations on shareholder rights, fraud investigations, and regulatory compliance, alongside writing and speaking on emerging technologies.

No summary available yet.

No summary available yet.

No summary available yet.

No summary available yet.

An advanced agent that perceives your screen and executes tasks by controlling the mouse, acting as a digital proxy to handle complex work on your behalf.

Amir Banifatemi is Chief Responsible AI Officer at Cognizant, where he leads the company’s global responsible AI strategy and governance across enterprise AI systems and platforms. A long-time AI entrepreneur, investor and innovation strategist, he has previously led the global AI XPRIZE and held leadership roles in initiatives such as GPAI and the OECD.AI expert community. Banifatemi serves as a board member of the International Association for Safe & Ethical AI (IASEAI) and advises international efforts on trustworthy, human-centric AI.

No summary available yet.

Senior Recruitment Specialist at Impact Ops and recruitment and operations professional with 15+ years of international experience across the nonprofit, development, and private sectors, supporting mission-driven organizations to build high-performing teams on global priorities.

No summary available yet.

No summary available yet.

No summary available yet.

I'm an experienced B2B tech content writer specializing in responsible tech.

Cortney Busch is Operations Director at Legal Advocates for Safe Science and Technology (LASST) and has worked in nonprofit operations for over 15 years. After obtaining her law degree and LL.M. from the City Law School, University of London, she has held operations roles in human rights and Effective Altruism organizations.

Rob Miles is an advisor to Nonlinear and a science communicator focused on AI safety and alignment. The Nonlinear team biography notes that he runs the Rob Miles AI YouTube channel and The Alignment Newsletter Podcast and collaborates with organizations such as MIRI, the Future of Humanity Institute, and the Centre for the Study of Existential Risk to help communicate their work.

Caroline Jeanmaire is an AI governance researcher and policy professional currently leading The Future Society's Washington, D.C. team, where she works with Congress and federal agencies on AI governance. She completed her DPhil in Public Policy at the University of Oxford's Blavatnik School of Government (2021-2024), where she researched models of international coordination to ensure the safety and reliability of AI systems under Professor Jonathan Wolff, supported by a Long-Term Future Fund grant. Prior to her doctorate, she served as Director of Strategic Research and Partnerships at UC Berkeley's Center for Human-Compatible AI (CHAI), where she built the research community around AI safety and managed external partnerships. Earlier, she was an AI Policy Researcher and Project Manager at The Future Society, a think-tank associated with Harvard Kennedy School, where she organized the first and second Global Governance of AI Forums at the World Government Summit in Dubai. She holds dual master's degrees in International Relations from Peking University and Sciences Po Paris, and a bachelor's degree in Political Sciences from Sciences Po Paris. She has been recognized as one of the "100 Brilliant Women in AI Ethics" and as a "35 under 35" future leader by the Barcelona Centre for International Affairs.

No summary available yet.

No summary available yet.

Justin is the Video Specialist at Giving What We Can, producing short‑ and long‑form videos for GWWC and its subsidiary channels. He studied English and Film at Stanford University, worked at McKinsey, and has also worked as a freelance food critic, published short fiction, and produced a political podcast.

Iván Arcuschin Moreno is an AI safety researcher based in London, UK, currently serving as Lead Research Scientist at Poseidon Research, a US-based AI safety non-profit where he heads the London research team. He holds a Computer Science PhD from the University of Buenos Aires, Argentina (2018–2024), where his thesis focused on automated test generation for Android apps. He completed two terms at the ML Alignment & Theory Scholars (MATS) program: the first (Jan–Jul 2024) under Adrià Garriga-Alonso at FAR AI, producing InterpBench, a collection of 86 semi-synthetic transformers with known circuits for evaluating mechanistic interpretability techniques (NeurIPS 2024); and the second (Jan 2025–Feb 2026) under Arthur Conmy at Google DeepMind, focusing on chain-of-thought faithfulness. He is lead author on "Chain-of-Thought Reasoning In The Wild Is Not Always Faithful" (ICLR 2025 workshop, 140+ citations within a year) and a contributor to the MIB mechanistic interpretability benchmark (ICML 2025). He also co-authored a position paper with Yoshua Bengio and founded AI Safety Argentina (AISAR), a research scholarship program to grow the AI safety research community in Latin America, supported by a $77,000 grant from Coefficient Giving.

Director of the FutureTech research group at MIT and principal research scientist at MIT CSAIL and the Initiative on the Digital Economy, whose work studies the economic and technical foundations of progress in computing and artificial intelligence.

12-month salary for researching value learning

David Udell is an independent AI alignment researcher and Content Manager at Iliad, an organization focused on applied mathematics research for AI alignment. He is based in Berkeley, CA. He participated in the SERI MATS program, where he worked on Team Shard's research on shard theory under mentors including Alex Turner (TurnTrout), and has since continued full-time alignment research. His research covers mechanistic interpretability, activation engineering, and alignment distillation: he co-authored work on steering language models via activation vectors, contributed to research on understanding and controlling maze-solving policy networks, and has worked on sparse circuit discovery for GPT-2-small. He has written extensively on LessWrong and the Alignment Forum, authoring a sequence of alignment distillations titled "Winding My Way Through Alignment" and numerous posts on shard theory, interpretability, and related topics. He has received multiple grants from the Long-Term Future Fund supporting his independent research. He is currently involved with the Iliad Fellowship and Iliad Intensive, programs offering mentored technical AI alignment research, and co-organized Agent Foundations 2026 at Carnegie Mellon University.

No summary available yet.

dumb in a fun way

General support of research led by David Lorrell

Jessica Rumbelow is the CEO and co-founder of Leap Laboratories (Leap Labs), an AI interpretability startup based in London, UK. She holds a PhD in model-agnostic interpretability from the University of St Andrews (2020–2023), where she also completed an MSc in Advanced Computer Science, and her doctoral research introduced novel techniques including Hierarchical Perturbation (HiPe) and the Proxy Model Test for evaluating saliency mapping algorithms. Prior to founding Leap Labs, she worked as a Research Scientist at the University of St Andrews applying deep learning to digital pathology, and held roles as Technical Alignment Research Scientist at Aligned AI and Research Scholar at SERI (Stanford Existential Risks Initiative). She participated in the MATS (ML Alignment Theory Scholars) Summer and Autumn 2022 cohorts, during which she co-authored the widely-cited "SolidGoldMagikarp" research on anomalous tokens in GPT-2 and GPT-3 with Matthew Watkins. She has been granted the title of Affiliated Lecturer by the Department of Computer Science and Technology at the University of Cambridge, and serves on the Advisory Board of the London Initiative for Safe AI (LISA). Leap Labs received seed funding from the AI Risk Mitigation Fund to develop its universal interpretability engine.