Sources

• Epoch AI (opens in a new tab)

  Forecasting

  Jaime Sevilla · Tamay Besiroglu (co-founder)

  Compute, capability, and buildout data; trusted across camps.

• SemiAnalysis (opens in a new tab)

  Infra analysis

  Dylan Patel

  Chips, datacenters, buildout economics.

• Stratechery (opens in a new tab)

  Business-model and platform-shift lens.

• AI Futures Project (opens in a new tab)

  Forecasting

  Daniel Kokotajlo · Eli Lifland · Thomas Larsen · Romeo Dean

  AI 2027, AI 2040: Plan A, quarterly timelines updates, AI Futures Model.

• Metaculus AI forecasts (opens in a new tab)

  Community aggregate estimates on AGI and related milestones.

• AI Safety at the Frontier (opens in a new tab)

  Best monthly digest of safety papers. Subscribe to stay current without reading arXiv.

• Dwarkesh Podcast (opens in a new tab)

  Highest signal density in the space; frontier labs treat it as peer review.

• The Cognitive Revolution (opens in a new tab)

  Fast on new capabilities, practitioner-focused.

• 80,000 Hours Podcast (opens in a new tab)

  Field-building

  Careers and podcast; canonical AI risk reading list.

• Machine Learning Street Talk (opens in a new tab)

  Technical depth, skeptic-leaning on AGI claims.

• No Priors (opens in a new tab)

  The VC/industry read; a deliberate counter-frame.

• Latent Space (opens in a new tab)

  For an LP who wants to see what builders actually deploy.

• Don't Worry About the Vase (opens in a new tab)

  Best weekly aggregator. One subscription to stay current.

• Transformer (opens in a new tab)

  Best AI-policy news feed; broke and covered the 2026 incidents well.

• METR (opens in a new tab)

  Technical safety

  Beth Barnes · Ajeya Cotra · Hjalmar Wijk

  Time-horizon evals, Frontier Risk Report, and the independent investigation of the OpenAI/HF incident. The de facto third-party auditor of frontier labs.

• Redwood Research (opens in a new tab)

  Technical safety

  Buck Shlegeris · Ryan Greenblatt

  AI control research; co-authored the HF incident investigation.

• Apollo Research (opens in a new tab)

  Technical safety

  Marius Hobbhahn

  Deception and scheming evals.

• FAR.AI (opens in a new tab)

  Technical safety

  Adam Gleave

  Adversarial robustness and safety research; runs field-building events.

• Palisade Research (opens in a new tab)

  Technical safety

  Jeffrey Ladish

  Offensive-capability and loss-of-control demonstrations.

• Alignment Research Center (opens in a new tab)

  Technical safety

  Paul Christiano (founder)

  Alignment theory; METR spun out of ARC Evals.

• Anthropic (opens in a new tab)

  Frontier lab

  Dario Amodei · Chris Olah · Jan Leike

  RSP, interpretability, sabotage risk reports, threat intel.

• OpenAI (opens in a new tab)

  Frontier lab

  Preparedness Framework; author of the HF incident technical report.

• Google DeepMind (opens in a new tab)

  Frontier lab

  Frontier Safety Framework; scheming-inability safety case.

• AI Digest (opens in a new tab)

  Forecasting

  Interactive explainers and annual forecasting surveys.

• Forecasting Research Institute (opens in a new tab)

  Forecasting

  Phil Tetlock (adviser)

  Rigorous forecasting; feeds the International AI Safety Report scenarios.

• Forethought (opens in a new tab)

  Strategy

  Will MacAskill · Fin Moorhouse

  Intelligence-explosion preparedness, broad timelines.

• Centre for the Governance of AI (opens in a new tab)

  Governance

  Oxford-rooted AI governance research and talent pipeline.

• Center for Security and Emerging Technology (opens in a new tab)

  Governance

  US policy research on compute, export controls, China.

• RAND (opens in a new tab)

  Governance

  Mainstream policy shop with a large AI program.

• IAPS (opens in a new tab)

  Governance

  Compute governance and international coordination.

• CNAS (opens in a new tab)

  Governance

  National-security policy on AI.

• CSIS

  Governance

  National-security policy on AI.

• Centre for Long-Term Resilience (opens in a new tab)

  Governance

  UK policy on extreme risks.

• UK AI Security Institute (opens in a new tab)

  Government

  Pre-deployment testing agreements with labs; secretariat for the Bengio report; published INC-2026-07-28-01 and open-weight cyber measurements.

• US Center for AI Standards and Innovation (opens in a new tab)

  Government

  US government standards and evaluation body.

• Constellation (opens in a new tab)

  Field-building

  Berkeley research hub for safety orgs.

• Lightcone Infrastructure (opens in a new tab)

  Field-building

  Oliver Habryka

  Where most safety research is first discussed.

• Coefficient Giving (opens in a new tab)

  Funding

  Largest philanthropic funder of AI safety.

• Center for AI Safety (opens in a new tab)

  Advocacy

  Dan Hendrycks

  Statement on AI Risk; benchmarks; policy.

• Center for Humane Technology (opens in a new tab)

  Advocacy

  Tristan Harris · Aza Raskin

  The AI Dilemma; public-facing societal-risk narrative.

• Future of Life Institute (opens in a new tab)

  Advocacy

  Max Tegmark

  AI Safety Index (lab grading); pause-letter lineage.

• ControlAI (opens in a new tab)

  Advocacy

  Legislative advocacy; youth-led (Encode).

• Encode (opens in a new tab)

  Advocacy

  Legislative advocacy; youth-led (Encode).

• MIRI (opens in a new tab)

  Theory

  Eliezer Yudkowsky · Nate Soares

  If Anyone Builds It, Everyone Dies; the stop-before-ASI position.

• LawZero (opens in a new tab)

  Research

  Yoshua Bengio

  Bengio's nonprofit; chair of the International AI Safety Report.

• Stanford HAI (opens in a new tab)

  Data

  AI Index (annual).

Nothing matches these filters. Clear filters