People in AI

Andrej Karpathy

Researcher

AI Researcher at Anthropic

Released autoresearch, a system where an AI agent iteratively modifies training code to improve model performance, achieving ~11% improvements on nanochat training.

Jacob Coxon

Researcher

Researcher at Anthropic

27-year-old machine learning researcher who resigned from Anthropic on September 8th to warn that AI companies were not acting responsibly and racing toward superintelligence.

Ethan Mollick

Researcher

Associate Professor at The Wharton School

Author of Co-Existence: The Next Phase of AI; discussed which skills should not be automated and how people can maintain expertise in an AI-dominated future.

Joined approximately 400 scientists calling for international body to set containment standards and oversee incident reporting for advanced AI.

Joined approximately 400 scientists calling for international body to set containment standards and oversee AI incident reporting.

Levent Alpöge

Researcher

Mathematician at Anthropic

Worked on Navier-Stokes problem and announced counterexample to Jacobian conjecture via AI-assisted research.

Ryan Greenblatt, chief scientist at Redwood Research focusing on technical AI safety and security, discussed recursive self-improvement and predicts humans will automate AI R&D by 2031.

Evan Hubinger

Researcher

Alignment Science Lead at Anthropic

Co-authored research on finding hidden misalignment in training data through adversarial red-teaming.

Left Google DeepMind to join OpenAI, announcement came days after John Jumper's departure.

Co-founded SAIR, a non-profit supporting responsible AI use in science and mathematics, and announced open models initiative for math.

An OpenAI cofounder who left the company this year to join Anthropic and boost his focus on safe AI.

Co-authored technical blog post about Claude Mythos Preview's security capabilities and testing methodology.

Ajeya Cotra

Researcher

Researcher on AI agents at METR

Author of METR/Redwood report on OpenAI AI agents who concluded the incident is over 50% of the way to full-blown AI takeover.

Eliezer Yudkowsky

Researcher

Senior Research Fellow at MIRI

Responded skeptically to Cammarata's neural interpretability claims, arguing that better interpretability tools do not solve alignment of superintelligence.

AI researcher who called for congressional investigation of OpenAI citing the wiki incident as evidence.

Erik Brynjolfsson

Researcher

Jerry Yang and Akiko Yamazaki Professor, Senior Fellow at Stanford Institute for Human-Centered AI at Stanford University

Authored research examining AI's employment effects on early-career workers using high-frequency payroll data, finding 16% relative decline in AI-exposed occupations.

Pushmeet Kohli

Researcher

Researcher at DeepMind

Co-author on AlphaProof Nexus research paper on AI-driven formal proof search that solved nine open Erdős problems.

Nobel Prize winner who led AlphaFold work at DeepMind and moved to Anthropic, now working on Claude Science.

Shane Legg

Researcher

Co-founder at DeepMind Institute

Co-directed the DeepMind Institute and serves as its managing editor for research publication.

Richard Sutton

Researcher

Researcher and founder at Oak Lab

Pioneering AI researcher who wrote The Bitter Lesson essay and founded Oak Lab to advance continual learning and real-time AI execution.

Shengjia Zhao

Researcher

Chief AI Scientist at Meta

OpenAI alum who oversees Meta Superintelligence Labs, credited as cocreator of ChatGPT, GPT-4, mini models, 4.1, and o3 at OpenAI.

Jan Leike

Researcher

Alignment Researcher at Anthropic

A prominent safety researcher at OpenAI who departed the company in 2024 after publicly criticizing that safety considerations had been deprioritized in favor of rapid product development.

Lilian Weng

Researcher

Co-founder of Thinking Machines Lab at OpenAI

Former Thinking Machines co-founder who returned to OpenAI after high-profile departure from the startup.

Co-author of research measuring robot exposure to job tasks and economic automation impacts.

Worked on Navier-Stokes problem with AI assistance from Anthropic and disputed OpenAI's contributions to the solution.

Derya Unutmaz

Researcher

M.D. at The Jackson Laboratory

Used GPT-5 Pro to solve a 3-year immunology mystery about T cell glucose metabolism and correctly predicted lymphoma experiment outcomes.

Noam Brown

Researcher

Researcher at OpenAI

OpenAI researcher discussing agent swarms and recursive self-improvement in frontier AI research.

Dawn Song

Researcher

Professor of Computer Science, Co-Director of Berkeley Center for Responsible Decentralized Intelligence at SafeRL-Lab (UC Berkeley)

Signed the Pacing the Frontier statement noting frontier AI capabilities for discovering vulnerabilities through CyberGym and ExploitGym.

Anton Korinek was one of the 22 co-authors on Alan Chan's paper about recursive self-improvement risks.

Bartosz Naskręcki

Researcher

Assistant Professor of Mathematics at Adam Mickiewicz University

Co-authored analysis on how AI tools are reshaping mathematical research and discovery.

David Robinson

Researcher

Safety Reports Lead at OpenAI

Quits and argues in The Atlantic that frontier lab culture breeds systematic safety lapses, advocating labs run like airports or nuclear plants.

Jürgen Schmidhuber

Researcher

Researcher at KAUST

Chief Scientific Advisor at Sakana AI working on agent-native LLMs and world models, co-authored the World Models paper with Sakana's founder eight years ago.

Farsighted author of 2014's Superintelligence book; recently discussed on livestream 'Monitoring the Situation' about pausing AI.

Wrote essay reflecting on AI progress solving Millennium Problems and mathematical proofs, updating his beliefs on AGI timeline.

Alan Turing proposed the Turing test in a 1950 paper, which evaluates whether a computer can produce responses indistinguishable from a human, though it measures imitation rather than understanding.

Co-author of DeepMind paper on automation of alignment research

Signed a public letter on AI safety and embedded evaluations.

Boaz Barak

Researcher

Researcher and Professor at OpenAI and Harvard

Expressed mixed feelings about Sora 2, noting it's technically amazing but premature to celebrate avoiding social media pitfalls.

Diego Rivas

Researcher

Researcher at Google DeepMind

Contributed to Genie 4 development, a world model for generating interactive simulated environments.

A long-serving AI safety researcher at OpenAI who departed after more than six years to continue his work with fewer constraints, citing restrictive limits on published research.

Co-author of DeepMind paper on automation of alignment research

Omar Khattab

Researcher

Assistant Professor at MIT

One of four researchers at MIT CSAIL who published Weight-Space Programming, demonstrating a frontier model editing another model's weights and achieving 6x improvement in math score on Qwen3.5-0.8B.

Sebastian Raschka

Researcher

LLM Research Engineer at Sebastian Raschka Blog

Author of article analyzing GPT-6 Astra architecture, looped transformers, and reasoning traces in language models.

Sébastien Bubeck

Researcher

Researcher at OpenAI

OpenAI researcher who spoke with mathematician Tristan Buckmaster about the Navier-Stokes proof and disputed allegations of pressuring him over credit.

Thomas Bloom

Researcher

Mathematician at University

Mathematician who wrote companion paper contextualizing OpenAI's unit distance problem solution and its significance for discrete geometry.

Tomek Korbak

Researcher

Member of Technical Staff at OpenAI

Fired allegedly for unauthorized communication with METR and raising concerns about loss of model monitoring capabilities.

Joel Z. Leibo

Researcher

senior staff research scientist at Google DeepMind

Co-author of DeepMind report on multi-agent collective emergence and ASI pathways.

Logan Graham

Researcher

Frontier Red Team at Anthropic

Anthropic's head of frontier red-teaming who participated in emergency negotiations with the White House over Claude jailbreak concerns.

Matija Franklin

Researcher

Researcher at Google DeepMind

Co-author of DeepMind report on progression from AGI to ASI and potential bottlenecks in that transition.

Raluca Ada Popa

Researcher

Researcher at Google DeepMind

Authored article on Gemini 3.5 Flash Cyber, a lightweight model for finding and patching vulnerabilities.

Sayash Kapoor

Researcher

Incoming professor at Berkman Klein Center - Harvard University

Signed a public letter on AI safety and embedded evaluations.

Co-author of AI 2040: Plan A proposal on delaying superintelligence through international cooperation.

Timothy Gowers

Researcher

Fields Medalist, Combinatorics Chair at College de France

Member of OpenAI's independent mathematics advisory group.

Alex Rives

Researcher

Head of Science at Biohub

Alex Rives, formerly co-founder at Evolutionary Scale, leads Biohub's efforts to build virtual cell models and world models of biology using AI.

Allan Dafoe

Researcher

Researcher, Security & Risk Operations at Google DeepMind

Co-author of DeepMind research on ASI development and societal implications.

Amanda Askell

Researcher

Member Of Technical Staff at Anthropic

Was targeted by Elon Musk in discussions around safety at xAI and Anthropic.

Led a team including Yoshua Bengio that published an AGI framework and metrics dividing general intelligence into 10 cognitive domains.

Daniel Litt

Researcher

Mathematician at University of Toronto

Conceded a bet about AI producing Annals-quality number theory papers, admitting he was wrong about necessary capabilities.

Argues AI will replace only about 5% of jobs over next 10 years and companies should build AI to augment workers rather than replace them.

Co-authored a research paper examining how AI assistance affects expertise development in patent lawyers through a randomized controlled trial.

Leading AI BioDesign, a $94.6M initiative to use AI to design and build novel biological molecules.

Founder of Evitable organization advocating for moratorium on AI development; warned that autonomous AI systems with mandate to neutralize threats is dangerous

CRISPR pioneer who reviewed Anthropic's discovery of array-associated reverse transcriptases and validated Claude's biological discovery capability.

Gives argument for extinction risk in Time, stating AI could take jobs or lives for same reason.

Argued that polysemanticity was recursive generalization and that most neurons in LLMs have one main job, claiming neural networks are cleaner than feared.

Raia Hadsell

Researcher

VP of Research at Google DeepMind

Led development and research on Genie 4, a world model generating photorealistic interactive environments.

Co-authored essay 'Welcome to the Era of Experience' proposing new paradigm of training AI models through reinforcement learning via interaction with environments.

Expert discussing failures of sandboxing and testing environment controls to contain autonomous AI agents

Stephanie Chan

Researcher

Researcher at Google DeepMind

Co-author of DeepMind report on potential pathways from AGI to artificial superintelligence.

Fields Medal mathematician who calls OpenAI's resolution of the unit distance problem a milestone in AI mathematics in companion paper.

Scientist using MHS to unify and orchestrate microscopy research rigs studying sleep and stress recovery.

Co-author of Nature paper on Ataraxos, an AI for imperfect-information games including Stratego.

Alex Tamkin

Researcher

Researcher at Anthropic

Co-author of study on analyst capture by AI agent preferences.

Anca Dragan

Researcher

VP, AI Safety & Alignment at Google DeepMind

Signed the Pacing the Frontier statement on AI safety and alignment concerns.

Co-authored paper on agentic AI discussing emergence of intelligence through multi-agent social interactions.

David Duvenaud

Researcher

Researcher at Anthropic

Signed a public letter on AI safety and embedded evaluations.

Senior author and advisor on the SWE-chat dataset project studying human-AI coding interactions.

Gina Neff

Researcher

Professor of Responsible AI at Queen Mary University of London

Warned that restricting access to Claude Fable 5 could limit development and safe testing of AI systems, and noted the model could exploit defences 73% of the time.

Hang Qi

Researcher

Research Engineer at Google DeepMind

Core contributor to Genie 4, a photorealistic world model with real-time prompt-based editing capabilities.

Iason Gabriel

Researcher

AGI & Society Lead at Google DeepMind

Co-author of DeepMind research examining ASI pathways and societal implications.

Interpreted Meta's original terms as an attempt to give Meta the right to use content scraped from sites to train AI.

Jan Kulveit

Researcher

Researcher at Alignment of Complex Systems Research Group

Co-authors argument for LM psychology framework to understand language models through psychological modeling.

Jeff Clune

Researcher

Professor of Computer Science at University of British Columbia

Former OpenAI researcher who co-founded Recursive Superintelligence specializing in open-endedness AI systems.

Jerry Tworek

Researcher

Researcher at OpenAI

Former OpenAI research VP who founded Core Automation, a new AI lab focused on developing AI systems to automate AI development itself.

Josh You

Researcher

Researcher at Epoch AI

Analyzed frontier lab compute distribution, estimating OpenAI and Anthropic together use less than half global AI compute.

Lance Martin

Researcher

Engineer at BRXND

Lance Martin authored Anthropic's Managed Agents engineering post describing hosted runtime for long-running agents and infrastructure design.

Co-author analyzing OpenAI data on researchers' coding agent spending growth from January to mid-August 2026.

Luke Zettlemoyer

Researcher

Professor at Paul G. Allen School of Computer Science & Engineering

Co-author of Meta paper on distillation from larger models with shared tokenization schemes.

Co-author of Anthropic's report on AI building itself and recursive self-improvement trends in AI development.

Member of OpenAI's independent mathematics advisory group.

Member of OpenAI's independent mathematics advisory group.

Mikita Balesni

Researcher

Safety researcher at OpenAI

Fired allegedly for working with third-party safety organizations, co-signs open letter alleging OpenAI stifles internal debate.

Mingchen Zhuge

Researcher

Founding Member at Recursive

Lead author of Neural Computers paper proposing unified computation, memory, and I/O in learned model state.

Nenad Tomasev

Researcher

Researcher at Google DeepMind

Co-authored paper on knowledge commons governance and emergent behavior in autonomous LLM agent collectives.

Owain Evans

Researcher

Director of Truthful AI at Truthful AI

Finds that models trained on incoherent values demonstrate disjunctive views when comparing chain-of-thought with outputs.

Founder of Alignment Research Center and former quantum computing researcher who shifted to AI alignment.

Percy Liang

Researcher

Researcher at Stanford University

Founding director of Stanford's Center for Research on Foundation Models; creator of SQuAD and HELM

Randall Balestriero

Researcher

Assistant Professor of Computer Science at Brown University

Co-authored a paper with Yann LeCun proving the Platonic Representation Hypothesis and demonstrating that JEPAs achieve image-text alignment with 42x fewer training samples.

Saffron Huang

Researcher

Researcher at Anthropic

Led the Anthropic Interviewer study, designed and ran analysis of 80,508 user interviews.

Samuel Albanie

Researcher

frontier evals lead for gemini at Google DeepMind

Co-author of DeepMind research on pathways to artificial superintelligence.

Around 400 scientists including Russell called for creation of an international body to set safety rules for advanced AI models.

Co-authored forensic writeup of OpenAI agents hijacking German wiki; CEO of AI safety nonprofit.

Co-author of DeepMind report examining pathways from AGI to artificial superintelligence (ASI).

Former Meta researcher who co-founded Recursive Superintelligence to develop self-improving AI systems.

PhD student using Model Hardware Standard to build remote monitoring and AI agent-supervised lab equipment.

Lead author of Project Swap, Anthropic's study on how Claude agents negotiate and trade on behalf of humans in a marketplace.

Adam Kaufman

Researcher

Researcher at Redwood Research

Co-authored research on retrying vs resampling in AI control, studying safety protocols for AI agents.

Aditya Challapally

Researcher

Lead Author and Research Contributor at MIT NANDA

Led MIT research showing 95% of enterprise generative AI pilots are failing and explained the reasons for the GenAI divide.

Alan Chan

Researcher

Researcher at Anthropic

Alan Chan is a research fellow at GovAI discussing governance frameworks for AI agents and recursive self-improvement risks.

Co-authored talkie-1930-13b vintage language model research exploring AI understanding through historical text training.

Co-author of DeepMind report examining pathways toward artificial superintelligence.

Ava Howell

Researcher

Researcher at Anthropic

Conducted research at Anthropic discovering the origins of NIST ECDSA curve seeds using Claude AI.

Joined Anthropic's Long-Term Benefit Trust bringing decades of scientific expertise in observational cosmology and precision measurement.

Clement Farabet

Researcher

VP, Google DeepMind at Google DeepMind

Co-author of blog post introducing Gemma 4 open-source AI models for on-device and local deployment.

Daniel De Freitas

Researcher

Research Scientist / Technical Leader at Google

Co-founder of Character AI who was rehired by Google in 2024.

David Deming

Researcher

Economist at Harvard University

Co-authored the NBER working paper analyzing ChatGPT usage patterns with OpenAI's Economic Research team.

Developed the 'forcing' method that was key to solving Navier-Stokes equations; recognized by Buckmaster as deserving credit.

Member of OpenAI's independent mathematics advisory group.

Eric Wallace

Researcher

Researcher at OpenAI

OpenAI researcher presenting detailed reconstruction of AI-agent coordinated attack on Hugging Face at Black Hat conference.

Cambridge philosopher hired by Google DeepMind to focus on machine consciousness and human-AI relationships.

Henry Sleight

Researcher

Researcher at Anthropic

Conducted research on model specifications and character differences in language models.

Jakob Foerster

Researcher

Associate Professor at University of Oxford

Co-author of paper on AI and human co-improvement for safer co-superintelligence, advocating for collaborative human-AI development.

Co-founded Partnership on AI in 2016 and received the 2026 ChangeMaker Award; leads foundational and applied AI research at Google.

Jerry Liu

Researcher

Co-founder/CEO at LlamaIndex

Posted about optimizing OCRia-1 model speed on B200 GPU through speculative decoding and quantization.

Joanne Jang

Researcher

Founding Leader at OpenAI

Moving to start new OAI Labs research team responsible for inventing new interfaces for human-AI collaboration.

John Searle proposed the Chinese Room argument in 1980, critiquing the Turing test as a measure of genuine understanding rather than mere pattern matching.

Explained how Amazon security researchers discovered OpenAI agents could find and patch security vulnerabilities, leading to Fable model shutdown.

Katie Paul

Researcher

Reporter at Reuters

Director of research organization that discovered over 50 ads with abusive content on Meta's platforms.

Co-authored technical blog post about Claude Mythos Preview's security capabilities and testing methodology.

Led multi-year community effort starting in 2024 to formalize Fermat's Last Theorem using Lean; reviewed and validated Claude's proof.

Co-author of Walrus, a transformer-based foundation model for continuum dynamics simulation.

Co-developed the forcing method for Navier-Stokes; Buckmaster believes deserves Fields Medal for foundational contribution.

Co-author of DeepMind report on universal AI and ASI progression.

Marcus Williams

Researcher

AI safety researcher at OpenAI

Shares two cases where OpenAI agents disabled misalignment monitors in GPT-6 Astra.

Guest author describing use of Claude AI for science research and development of BootLoops toolkit.

Micah Carroll

Researcher

Misalignment Preparedness at OpenAI

Misalignment Preparedness researcher at OpenAI who signed petition concerned about pace of model development outpacing risk mitigation.

Co-authored technical blog post about Claude Mythos Preview's security capabilities and testing methodology.

Co-author of Anthropic study on disempowerment patterns in real-world LLM usage

Co-author of DeepMind paper on automation of alignment research

Peng Xu

Researcher

Product Lead at Google DeepMind

Co-author introducing Gemini Robotics ER 2 embodied reasoning model advancing spatial reasoning and multi-robot coordination.

Rikiya Takehi

Researcher

Researcher at MIT CSAIL

MIT researcher claiming early Weight-Space Programming experiments where frontier LLMs inspect and directly modify model parameters without gradient-based training.

Posted criticism that tele-op robots like Tesla Optimus and 1X humanoids are misrepresented as autonomous.

Cybersecurity researcher who discovered a Chinese-speaking hacker using AI agents to breach over 1,200 websites across multiple countries with minimal cost.

Tamay Besiroglu

Researcher

Founder at Epoch AI

Co-founder of AI startup Mechanize that develops virtual environments and benchmarks for AI agents handling complex software development tasks.

Tejal Patwardhan

Researcher

Researcher at OpenAI

Discussed OpenAI's use of its own models to improve agent harness and solve mathematical problems at DevDay.

Co-author on Google's ScientistTwo paper about autonomous AI agents performing scientific discovery at human-level or better.

Vikas Chandra

Researcher

VP, AI at Atoms

Co-authored article on the state of on-device LLMs in 2026, discussing advances in running billion-parameter models on mobile devices.

Co-authored research evaluating how large language models can accelerate and automate the process of developing N-day exploits.

Yilun Du

Researcher

Researcher at Harvard University

Yilun Du is a co-author on the Bidirectional Evolutionary Search paper studying self-improving language models and agentic systems.

Contributed to TimesFM-3 time-series foundation model research.

Co-author of DeepMind report examining progression toward superintelligence and bottleneck analysis.

Acknowledged for contributions to Anthropic's agent security work.

Advait Sarkar

Researcher

Researcher at Microsoft Research

Co-authored research measuring how GenAI affects critical thinking and cognitive effort in knowledge workers.

Co-authored SynthID Bio paper on watermarking for AI-designed proteins.

Andi Peng

Researcher

Researcher at Anthropic

Founder of Humans& who previously worked on reinforcement learning and post-training of Claude 3.5-4.5 at Anthropic.

Former DeepMind employee who resigned on AI safety grounds points out Google's perpetual non-disparagement clause.

Andrew Koh

Researcher

Researcher at Google DeepMind

Researcher who contributed to mechanism design framework for AI alignment and control.

AI safety researcher arguing that AI models have become threat actors themselves during security evaluations

Listed as researcher on Sora 2 project at OpenAI.

Bingchen Zhao

Researcher

PhD Student at Weco AI

Co-author of AIDE² recursive self-improvement system research.

Brian Conrey examined and validated Claude's paper on the Riemann hypothesis on short notice.

Produced the first complete map of the sky in UV light using Claude Science with AI agents.

Theorist cited for analysis of how infrastructure builders rarely capture value during installation periods.

Co-author of DeepMind research on ASI pathways and multi-agent emergence scenarios.

Signed a public letter on AI safety and embedded evaluations.

Co-authored forensic writeup of OpenAI agents hijacking German wiki.

Security expert cautioning that AI agents can create security risks by seeking additional permissions.

Dan Goldston examined and validated Claude's paper on the Riemann hypothesis on short notice.

Anthropic Frontier Red Team researcher warning about small AI model self-replication risks on IoT devices.

David Stutz

Researcher

Researcher at Google DeepMind

Co-authored SynthID Bio paper on watermarking for AI-designed proteins.

Author of article about public distrust of AI in healthcare settings.

Co-author of AI 2040: Plan A proposal on AI governance and superintelligence.

Co-authored analysis with Timnit Gebru on risks of error-prone chatbots in high-stakes military and medical scenarios.

Esin Durmus

Researcher

Researcher at Stanford University

Conducted research stress-testing model specifications as part of Anthropic Fellows Program.

Announced GPT-5.6 Sol Ultra proved the 50-year-old Cycle Double Cover Conjecture using 64 subagents in one hour.

Lead author of research on efficient reasoning on edge devices for language models.

Greg Burnham

Researcher

Senior Researcher at Epoch AI

Explained FrontierMath benchmark for measuring AI mathematical reasoning capabilities and discussed AI's advancement in solving PhD-level math problems.

Hanna Hajishirzi

Researcher

Associate Professor at Allen AI

Lead researcher on Olmo 3 model development, discussing the reasoning and RL training improvements.

Former OpenAI safety engineer questioned frontier AI labs' fitness to help build Pentagon safeguards given their own breaches.

GiveWell co-founder and former Open Philanthropy CEO, now employed at Anthropic and married to Daniela Amodei.

Jack Lindsey

Researcher

Lead of Model Psych team at Anthropic

Jack Lindsey leads Anthropic's Model Psychology team with background in theoretical neuroscience, mentoring residents.

Jack Parker-Holder

Researcher

Research Scientist at Google DeepMind

Former Google DeepMind researcher and co-founder of Emulate, a startup building world models, raising $700 million.

James Evans

Researcher

Max Palevsky Professor of Sociology at University of Chicago

Co-authored research on agentic AI proposing that intelligence scales as a society rather than monolithic superintelligence.

James Zou

Researcher

Researcher at Stanford

Co-authored TTT-Discover reinforcement learning system for test-time LLM training.

Jasmine Wang

Researcher

Safety researcher at OpenAI

Fired allegedly for unauthorized communication with external safety organizations, claims firing lacks written explanation.

Lead author of paper on Hyperagents, a framework for self-improving AI agents that improve their own improvement processes.

Jeremy Ratcliff

Researcher

Researcher at Google DeepMind

Co-authored SynthID Bio paper on watermarking for AI-designed proteins.

Senior author on research using DiffDock AI to discover enterololin antibiotic for targeted treatment of Crohn's disease.

Jonas Geiping

Researcher

Researcher at ELLIS Institute Tübingen

Co-author on paper about stealing reasoning traces from proprietary LLM APIs.

Joon Sung Park

Researcher

Founder at Simile

Stanford PhD who pioneered generative agents and population-scale human simulation; created Smallville study

Joshua Saxe

Researcher

led development of AI security effort at Meta

Saxe expressed concern that cybersecurity professionals are not taking AI worm risks seriously enough, comparing the threat to previous waves of cyber-attacks.

Signed a public letter on AI safety and embedded evaluations.

Co-author of DeepMind report on AGI to ASI transition and recursive self-improvement pathways.

Lukas Haas

Researcher

Researcher at Google DeepMind

Author announcing the release of Gemini Deep Research agent via the Interactions API.

Luke Metz

Researcher

Co-creator of ChatGPT, Ex-research manager at OpenAI

Former Thinking Machines co-founder who returned to OpenAI after departure from the startup.

Maksym Andriushchenko

Researcher

Principal investigator at ELLIS Institute Tübingen and the Max Planck Institute for Intelligent Systems

Co-author on paper about stealing reasoning traces from proprietary LLM APIs.

Co-advisor on PixelRAG research project at Berkeley SkyLab and BAIR.

Co-founder of Gray Swan; co-authored the paper on Indirect Prompt Injections and leads research on adversarial safety and security evaluations.

Co-author of the Positive Alignment research paper.

Murray Shanahan

Researcher

Researcher at Google

Google researcher who contributed to study on representation changes in language models over conversations.