// category · 9 tools · 0 open source
AI Research Labs
Research labs, nonprofits and government institutes studying advanced AI, its trajectory and its risks. Their publications are primary sources for anyone tracking the frontier.
Who these organizations are
This category lists the labs that build frontier models and the independent groups that test them, track progress and study safety. They matter to anyone following AI because their reports are often the primary source behind news about model risks and capabilities.
There are three broad kinds:
- Frontier labs such as Google DeepMind build the models themselves and publish safety frameworks. OpenAI's Preparedness Framework and Anthropic's Responsible Scaling Policy set the capability thresholds that trigger extra safeguards; our Preparedness Framework vs Responsible Scaling Policy guide compares them.
- Government institutes such as the UK AI Security Institute test models before and after release and fund outside research.
- Independent research groups such as METR, Epoch AI, Apollo Research, Redwood Research, the Center for AI Safety and MIRI evaluate dangerous capabilities, measure compute and benchmark trends, or work on control and alignment. The AI Alignment Forum is where much of this research is published and debated.
How to use these pages: when a lab announces a new model, check whether METR or a government institute published an evaluation, and read the lab's own system card. When someone claims AGI or superintelligence has arrived, compare it with how these groups define and measure it.
Related guides: has AGI been achieved, superintelligence vs AGI, and our explainer on the superintelligence executive order, EO 14434, which changed how the US government names AI but not how researchers define it.
METRmetr.orgNonprofit that evaluates frontier models for dangerous autonomous capabilities and measures how long a task AI agents can complete.Labs & InstitutesFreeHN 18
UK AI Security Instituteaisi.gov.ukUK government body that tests frontier models before and after release and funds research into AI security.Labs & InstitutesFreeHN 8
MIRIintelligence.orgThe Machine Intelligence Research Institute, one of the earliest organisations focused on risks from smarter than human AI.Labs & InstitutesFreeHN 3
Redwood Researchredwoodresearch.orgNonprofit research lab working on AI control: techniques for safely using powerful models even if they are misaligned.Labs & InstitutesFreeHN 3
Alignment Forumalignmentforum.orgOnline hub where alignment researchers publish and debate technical work on making advanced AI safe.Labs & InstitutesFree
Apollo Researchapolloresearch.aiAI safety organisation studying deceptive behaviour and scheming in advanced models through evaluations and interpretability.Labs & InstitutesFree
Center for AI Safetysafe.aiNonprofit doing technical safety research, field building and compute support for researchers studying risks from advanced AI.Labs & InstitutesFree
Epoch AIepoch.aiResearch institute that tracks trends in AI compute, data, algorithms and benchmarks, publishing open datasets and analysis.Labs & InstitutesFree
Google DeepMinddeepmind.googleGoogle's AI research lab behind Gemini, AlphaFold and many landmark results in reinforcement learning and science.Labs & InstitutesSee site// guides
Guides to AI research labs
How to Get Into Alignment: A 2026 Roadmap
How to get into alignment in 2026: the skills you need, free AI safety courses, paid fellowships like MATS and Anthropic Fellows, career grants and first jobs.
What Is AI Safety? Risks, Research and Who Does It
What is AI safety? A clear 2026 explainer: the three kinds of AI risk, how safety differs from alignment and security, the main research areas and who does it.
Superintelligence vs AGI: What's the Difference?
Superintelligence vs AGI: AGI is roughly human level general intelligence; superintelligence far exceeds it. Lab definitions, DeepMind levels, and US SI.
What Does AI21 Labs Do in 2026? A Clear Answer
What does AI21 Labs do? The Israeli lab behind Jurassic and Jamba now trains custom models for companies. Its history, products, pivot and what it means for you.
OpenAI Sandbox Escape: Why Training Went Offline
The OpenAI sandbox escape explained: how a research model used DNS to reach the internet on Sept 20, why training paused, and what offline training changes.
What Is Meta Superintelligence? MSL Explained
What is superintelligence at Meta? It's Meta Superintelligence Labs (MSL), the unit behind Muse, plus Zuckerberg's goal of personal superintelligence.
EO 14434 Explained: Super Intelligence Executive Order
EO 14434, signed September 29, 2026, tells federal agencies to say 'SI' instead of 'AI'. What it requires, what stays the same, and the November 28 deadline.
Preparedness Framework vs Responsible Scaling Policy
Preparedness Framework vs Responsible Scaling Policy: how OpenAI and Anthropic set AI risk thresholds, who approves releases, and how Google's version compares.
Has AGI Been Achieved? GPT-6 Astra and the Debate
Has AGI been achieved? What OpenAI really said about GPT-6 Astra, how AGI is defined, the evidence on each side, and what a formal declaration would change.
UK AI Security Institute: What It Does in 2026
The UK AI Security Institute explained: who runs it, how it tests frontier models before release, what the 2026 Mythos 5 incident showed and how to use it.
AGI vs LLM: What's the Difference in 2026?
AGI vs LLM explained: an LLM is a kind of AI model that predicts text, while AGI is a goal for human level general ability. See where today's chatbots sit.
What Does SI Mean Instead of AI? Super Intelligence
What does SI mean instead of AI? Under EO 14434, US agencies now say Super Intelligence. What SI covers legally and how it differs from superintelligence.
Trump AI Executive Orders: 2025 to 2026 Timeline
Trump AI executive orders and key US AI policy dates, January 2025 to October 2026: the AI Action Plan, Genesis Mission and EO 14434, with official links.
Pace the Frontier: Dario Amodei's Plan Explained
Pace the frontier explained: Dario Amodei's three step plan to slow AI capability gains, why he proposed it, how OpenAI and xAI reacted, and what changes next.
Genesis Mission Explained: The US AI for Science Plan
The Genesis Mission explained: what EO 14363 set up, the DOE platform and deadlines, the $5 billion July 2026 expansion, the 278 projects, and how to take part.
Claude Fable vs Opus vs Mythos: Which Claude to Use
Claude Fable vs Opus vs Mythos, plus Mythos vs Opus 5.5: how Fable 5.1, Opus 5.5, Sonnet 5.5 and Mythos differ in price, speed, safeguards and access.
Gemini 4 Argon Release Date: Now Available
Gemini 4 Argon release date is October 13, 2026. Who can use it in the Gemini app, API, and AI Studio, intro pricing through November 30, and Fairwind access.
National AI Policy Framework: SI Framework Explained
The White House National AI Policy Framework, now called the SI framework, explained: its seven sections, state preemption, what Congress must do, and status.