Sources

• Environmental Disclosure Index MIT

  This site Updated quarterly

  Our own reading of what frontier developers and the operators of their data centres have published about training energy, emissions, power sources and energy per prompt, with permits and press reporting marked separately from company statements. Most cells record an absence. Company figures are self-reported and mostly unaudited, and market-based emissions reflect contracts rather than the grid at the hour of use.

  • datacentre-electricity 7
  • datacentre-shares 6
  • environment-developers 7
  • grid-carbon-free 22
  • inference-energy 3
  • operator-energy 4
  • training-footprints 9

  Reuse permitted.

• Notable AI Models CC-BY-4.0

  Epoch AI Updated weekly

  Inclusion is editorial ("notability criteria"), not exhaustive. Compute figures are frequently estimates; check the per-row confidence and notes columns before treating any value as measured.

  • compute-thresholds 51
  • frontier-models 540
  • model-releases 1,083
  • release-disclosures 79

  Reuse permitted.

• Use of artificial intelligence in enterprises EC-reuse-2011-833-EU

  Eurostat Updated annual

  Self-reported by enterprises in the ICT usage survey, so it measures what firms say they use and what they recognise as AI. Covers the EU plus a few reporting non-members, and only enterprises with ten or more people employed unless the size dimension says otherwise. Agriculture, forestry, fishing, mining and the financial sector are outside the survey population entirely. Small cells are suppressed rather than zero, and NACE activity classes nest, so shares by industry do not partition anything.

  • enterprise-ai-country 33
  • enterprise-ai-industry 50
  • enterprise-ai-size 3
  • enterprise-ai-technology 7

  Reuse permitted, attribution required.

• AI Usage Disclosure Index MIT

  This site Updated quarterly

  Our own compilation of the token volumes, user counts and shares of use that companies and studies have put on the record, each linked to where it was stated. Scopes, units and token definitions differ between companies, so no two figures are comparable without the scope printed beside each. Figures known only through press reporting are marked as such. Not a market share estimate, and not a measure of any company's actual volume beyond what it chose to state.

  • token-volumes 21
  • usage-purposes 14
  • usage-reach 9

  Reuse permitted.

• AI Benchmarking Hub CC-BY-4.0

  Epoch AI Updated weekly

  Mixes Epoch's own evaluation runs with externally reported scores; these are not strictly comparable. Benchmark questions themselves remain the property of their creators and are not redistributed here. External rows retain their original licensing.

  • benchmark-frontier 18

  Reuse permitted.

• AI Data Centers CC-BY-4.0

  Epoch AI Updated several times a week

  The largest AI sites Epoch has found, mostly in the United States. Epoch estimated in September 2026 that they hold 44% of the world's deployed AI compute, with a 90% interval of 23% to 81%. Power, compute and cost are Epoch's estimates from satellite imagery, permits and company statements; four in five IT power estimates fall within a factor of 1.4 of the true value. The owner is the owner of the chips, not necessarily of the building. Capacity after the date of retrieval is Epoch's projection. Coordinates are read from the hub's map page, which publishes them; the downloads do not.

  • ai-datacentres 95

  Reuse permitted.

• AI Datacentre Index MIT

  This site Updated quarterly

  AI datacentre sites that Epoch AI's hub does not hold, recorded from the statements of developers, named tenants and planning authorities. Australia only so far. A site is included only where a primary source says it is built for AI compute or names the AI deployment inside it. Capacity is as the developer states it, IT or facility power as labelled, and is not an estimate; most of it is proposed and not yet approved. Coordinates are this site's placement of each stated address or area.

  • datacentre-index 13

  Reuse permitted.

• AI Diffusion Report: AI User Share MIT

  Microsoft AI for Good Lab Updated quarterly

  Derived from anonymised Microsoft telemetry, then adjusted for OS/device market share, internet penetration and a modelled mobile-to-desktop ratio. It therefore measures Microsoft-visible AI use, scaled, not observed total AI use. Systematically less reliable where Microsoft platform share is low, and it cannot see AI embedded in products users do not recognise as AI.

  • ai-exposure 147

  Reuse permitted.

• AI Governance Readiness Index MIT

  This site Updated quarterly

  Our own editorial coding from primary legal sources, not an objective measurement. Every score cites a specific instrument and article. "Readiness" means the machinery exists to govern AI. It is emphatically not a judgement that the resulting rules are good, proportionate, or well enforced.

  • governance-readiness 58

  Reuse permitted.

• AI Incident Database ODbL-1.0

  Responsible AI Collaborative Updated continuous

  ODbL share-alike: we display incident headlines with a link back, which is display, and must NOT publish a derived database dump without licensing it alike. Incidents are reported, not sampled, so coverage reflects media attention rather than incidence, and an absence of incidents in a country usually means an absence of reporting.

  • news-feed 113

  Reuse permitted, derived databases must be licensed alike.

• AI Law and Policy Index MIT

  This site Updated quarterly

  Our own reading of primary instruments, with a link to each. Not legal advice, not a compliance checklist and not exhaustive. Instruments that never mention AI are included where they apply to it, which is most of the Australian list. Statuses and commencement dates go stale; every record carries its review date.

  • policy-index 86

  Reuse permitted.

• arXiv cs.AI new submissions CC0-1.0

  arXiv Updated daily

  arXiv offers its metadata under CC0. Abstracts remain the authors' work, so we cap the extract at the same 200 characters as every other feed here and always link to the paper. Preprints are not peer reviewed. Volume is high and inclusion in this feed means the day's submissions matched our keyword filter, not that anyone judged the work important.

  • news-feed 113

  Reuse permitted.

• arXiv cs.CY new submissions CC0-1.0

  arXiv Updated daily

  Computers and Society: where most arXiv work on AI policy, governance and societal impact is filed. Same terms and same caveats as cs.AI. Preprints are not peer reviewed.

  • news-feed 113

  Reuse permitted.

• Canada news, Innovation Science and Economic Development OGL-Canada-2.0

  Government of Canada Updated daily

  Departmental news releases are political statements about policy, not neutral descriptions of it. The Open Government Licence forbids suggesting the Government of Canada endorses this site, which it does not.

  • news-feed 113

  Reuse permitted, attribution required.

• Compute Threshold Index MIT

  This site Updated quarterly

  Training-compute thresholds written into law, executive action or published policy, with the text that sets each one. Whether a given model crosses a threshold here is arithmetic on Epoch's compute estimates, most of which are estimates rather than disclosures, and is not a legal finding about any model or developer.

  • legal-thresholds 6

  Reuse permitted.

• Federal Register API public-domain

  US Office of the Federal Register Updated daily

  Full-text keyword matching on "artificial intelligence" returns documents that merely mention AI in passing. Requires topical filtering before display.

  • news-feed 113

  Reuse permitted.

• Frontier Safety Framework Index MIT

  This site Updated quarterly

  Our own reading of each developer's published safety framework and model cards. Every field is what the developer says about itself: these are self-assessments under self-written frameworks with no common scale and no external audit, and they are indexed here for exactly that reason. Not a ranking, and not a verification.

  • safety-frameworks 8

  Reuse permitted.

• GOV.UK news and communications OGL-3.0

  UK Government Updated daily

  Keyword-filtered search feed, so relevance varies. Departmental press releases are political communications, not neutral descriptions of policy.

  • news-feed 113

  Reuse permitted.

• GPU Clusters CC-BY-4.0

  Epoch AI Updated monthly

  Compiled from public reporting; each row carries a certainty flag ranging from confirmed to speculative. Chinese clusters are systematically less well documented than US ones, so country totals understate opacity as much as they measure capacity.

  • compute-clusters 36

  Reuse permitted.

• NIST news public-domain

  US National Institute of Standards and Technology Updated daily

  Covers all of NIST's remit; AI is a small fraction. Requires topical filtering.

  • news-feed 113

  Reuse permitted.

• OpenAlex CC0-1.0

  OurResearch Updated continuous

  An open index of scholarly works. Counts here are matches on a stated search phrase in title or abstract, not a curated field classification: a paper that studies alignment without using the phrase is missed, and one that mentions it in passing is counted. The exact query behind every number is printed with it. Recent years undercount because indexing lags publication, and the current year is always partial.

  • research-volume 6

  Reuse permitted.

• Release Safety Disclosure Index MIT

  This site Updated monthly

  Our own reading of the system card, model card or safety report each developer published for each release Epoch AI lists from 2025, joined to Epoch's list. Every cell is what a developer says about its own model under its own framework: whether a card came out with the release, whether it reports dangerous-capability results, which government institutes it names as pre-release testers, and whether it says a risk threshold was reached. None of it is verified. A release Epoch adds after the last review is shown as awaiting review.

  • release-disclosures 79

  Reuse permitted.

• Shaping Europe's digital future EC-reuse-2011-833-EU

  European Commission Updated daily

  Covers all EU digital policy, not only AI; requires topical filtering. Reuse decision does not extend to third-party material or EU logos.

  • news-feed 113

  Reuse permitted, attribution required.

• Artificial Analysis proprietary

  Artificial Analysis

  RULED OUT. The free API tier is internal-use only and forbids redistribution. Registered here so the build fails loudly if anyone ever wires it into a rendered series. Link out to them instead.

  Redistribution forbidden. Blocked in code.

• Task-completion time horizons of frontier AI models no-licence-stated

  METR

  RULED OUT for charting. METR's analysis repository carries no LICENSE file, and its README points at one that is not there, so no reuse permission has been granted and the default is reserved. Their published finding is referenced as an attributed statement with a link, which is a fact rather than their data. Re-plotting their measurements here would need permission we do not have.

  Redistribution forbidden. Blocked in code.

• LLM Rankings no-licence-verified

  OpenRouter

  RULED OUT for charting. The only public per-model token counts, measured on traffic routed through OpenRouter's own service, which is a developer-heavy sample rather than the market. No permission to reuse the data could be verified, so the default is reserved and nothing from it is plotted here.

  Redistribution forbidden. Blocked in code.

• AI Index Report CC-BY-NC-ND-4.0

  Stanford HAI

  RULED OUT for charting. No-derivatives means we may not re-plot their figures or restyle their data. Cite, link, and quote sparingly only.

  No derivative works. Cannot be re-plotted. Blocked in code.

The check runs at both ends. guard() in the Python pipeline refuses to write a dataset whose source forbids republication, and assertRenderable() in the site refuses to render one. Two checks rather than one, because data can also arrive hand-written without passing through the pipeline, which on this site it frequently does.

Licensed MIT, and under the second licence where a row names one. A CSV file holds the records only; its licences and citation are those of the JSON beside it. Lists within a record are joined by semicolons.