Data Sources
Every source behind our AI charts: who collected the data, which topic pages use it, and the licence or terms it comes with.
Data sources · 80 sources · 316 charts
Every number on AI Adoption comes from someone else's work: a research lab, a survey, a government agency. Here is whose, how we credit it, and the terms each one comes with.
- Main source
- Stanford AI Index 2026
- Sources
- 80 in 6 groups
- Open data
- 11 sources
- Watched weekly
- 8 sources
- Last reviewed
- 20 Oct 2026
301of our 316 charts are redrawn from the AI Index 2026 data filesSee the source →15charts are rebuilt from Epoch AI's newer data, with the download dateSee the source →11sources publish open data that anyone can reuse with creditBrowse →
Every chart, by where its data comes from
316 squares · one per chart · hover to see it, click to open ittap to see it, tap again to open it
Most-used sources · charts
Terms behind the 80 sources
Open data11Published datasets with an open licence or public-domain terms. We can rebuild and share charts from them.
Free to read59Reports, papers and leaderboards that anyone can read. We quote their figures and credit them.
Proprietary10Paid or private datasets. We use only the figures the AI Index published.
Our main source
Stanford HAI AI Index Report 2026
Stanford HAI's annual report brings together data from research groups, companies, pollsters and governments. We redraw each chart from its published data files and credit the organisation that collected the data. We do not copy the report's chart images or text.
Show the full citation
Sha Sajadieh, Loredana Fattorini, Raymond Perrault, Yolanda Gil, Vanessa Parli, Lapo Santarlasci, Juan Pava, Nestor Maslej, Russ Altman, Erik Brynjolfsson, Carla Brodley, Jack Clark, Virginia Dignum, Vipin Kumar, James Landay, Terah Lyons, James Manyika, Juan Carlos Niebles, Yoav Shoham, Elham Tabassi, Russell Wald, Toby Walsh, Dan Weld. “The AI Index 2026 Annual Report,” AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, Stanford, CA, April 2026. https://doi.org/10.48550/arXiv.2606.15708
- Charts
- 301 of 316
- Licence
- CC BY-ND 4.0
- Updates
- Annual (April)
- Data
- Files per figure
How we credit sources
Report a credit error →We name the source
Every chart names who collected the data, the year, and the AI Index figure it comes from.
Source: Baidu, 2025, via the AI Index Report 2026 (Stanford HAI), figure 2.7.7. Chart and text by AI Adoption.
We redraw, never copy
We draw every chart ourselves from the published numbers. When we rebuild one from a source's own data, the chart links that dataset, states its licence and gives the download date.
We open the data
Each chart page links to its data as a CSV file and gives a citation you can copy. Licences stay with each source: open data can be reused on its licence's terms; other figures are quoted for reference.
Showing all 80 sources
No source matches that search. Try a shorter word.
01
Data partners
Organisations that collect and maintain the datasets the AI Index draws on.
Epoch AI
Independent research group's open databases of notable AI models, training compute, AI hardware, data centres, supercomputers and benchmark results.
27 charts
Open data under CC BY 4.0. 15 of these charts are rebuilt from the source's own data; the rest use figures published in the AI Index 2026.
See the 27 charts
McKinsey & Company (QuantumBlack, AI by McKinsey)
Annual online survey of business leaders on how organisations adopt AI, manage its risks and see results; 1,993 respondents in 105 countries in 2025.
ProprietaryUpdates: annual
19 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 19 charts
Commercial market-intelligence platform whose company and funding data (from Capital IQ and Crunchbase) underpin the AI Index's private-investment figures.
ProprietaryUpdates: continuous
15 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 15 charts
Lightcast
Labour-market analytics firm that collects online job postings from over 51,000 sites and tags them with a taxonomy of more than 33,000 skills.
ProprietaryUpdates: continuous
13 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 13 charts
Code.org (with CSTA and ECEP Alliance)
Annual report on access to, and enrolment in, computer science in US public high schools, with AP computer science exam data from the College Board.
Free to readUpdates: annual
12 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 12 charts
LinkedIn
LinkedIn's research unit, which turns anonymised, aggregated member profiles into measures of AI talent, hiring, skills and migration by country.
ProprietaryUpdates: continuous
11 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 11 charts
Zeki Data
Talent-intelligence company tracking about 658,000 top AI researchers and inventors outside China across 21 countries, 2010–2025.
Proprietary
8 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 8 charts
International Federation of Robotics (IFR)
Industry body's annual statistics on industrial and service robot installations and operational stock worldwide.
ProprietaryUpdates: annual
7 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 7 charts
GitHub (with Harvard Business School and Microsoft Research)
GitHub's count of public AI repositories and their stars, mapped to countries by the owners' IP-address locations.
Proprietary
5 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 5 charts
Hugging Face (hub-stats dataset maintained by Caleb Fahlgren)
Regularly refreshed snapshot of public models and datasets on the Hugging Face Hub, used for counts of models and datasets over time.
3 charts
Open data under Apache-2.0. We use the figures as published in the AI Index Report 2026.
See the 3 charts
Cloudscene
Commercial directory of data centres, cloud and network providers, used for counts of data centres by country.
ProprietaryUpdates: continuous
2 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Digital Policy Alert (St. Gallen Endowment for Prosperity through Trade)
Independent tracker of digital-economy policy; its automated pipeline identified AI laws passed in G20 countries for the AI Index.
ProprietaryUpdates: continuous
2 charts
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Responsible AI Collaborative
Open, crowd-sourced database of real-world harms and near-harms caused by AI systems, built from news and other reports.
Open data · CC BY-SA 4.0 (core collections)Updates: continuous
1 chart
Open data under CC BY-SA 4.0 (core collections). We use the figures as published in the AI Index Report 2026.
See the 1 chart
Washington think tank that supplied the AI Index with data on AI application funding by country and application area.
Proprietary
1 chart
Proprietary data, not public. We use the figures as published in the AI Index Report 2026.
See the 1 chart
02
Surveys and polls
Polls of the public, workers, students and business leaders. Each chart states who was asked and when.
Ipsos
Annual online survey of adults in around 30 countries on how they feel about AI products, jobs, trust and regulation.
Free to readUpdates: annual
9 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 9 charts
Pew Research Center
Nonpartisan polling of US adults, AI experts and publics abroad on AI's effects on society, jobs and trust in regulators.
Free to readUpdates: irregular
6 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 6 charts
Computing Research Association (CRA)
Annual survey of PhD-granting computing departments in the US and Canada on degrees, enrolment, faculty and new PhDs' employment.
Free to readUpdates: annual
4 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 4 charts
The Civic Health and Institutions Project (CHIP50), with the COVID States Project
Large 50-state US survey on AI use, expected job impact and support for federal AI regulation, reported by state.
Free to readUpdates: periodic
3 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 3 charts
Forecasting Research Institute
Three-year panel tracking forecasts by AI experts, economists and the public on AI capabilities, adoption and social impact.
Free to readUpdates: periodic
3 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 3 charts
Chegg.org (Chegg, Inc.)
Online survey of university students in many countries on study habits, wellbeing and use of generative AI for coursework.
Free to readUpdates: periodic
2 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Ipsos for Google
Multi-country survey of adults, commissioned by Google, on AI use, jobs, companionship and regulation priorities.
Free to readUpdates: irregular
2 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 2 charts
University of Melbourne and KPMG International (Gillespie, Lockey, Ward, Macdade, Hassed)
Multi-country survey on public trust in AI, use of AI at work and AI literacy.
Free to readUpdates: irregular
2 charts
Free to read; published under CC BY-NC-SA 4.0. We use the figures as published in the AI Index Report 2026.
See the 2 charts
College Board research brief, based on repeated surveys from 2024 to 2025, on how US high school students use generative AI for schoolwork.
Free to readUpdates: periodic research briefs
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Imagining the Digital Future Center, Elon University (Rainie and Anderson)
US public opinion survey, compared with an expert canvass, on how AI may change human capacities such as empathy and critical thinking by 2035.
Free to readUpdates: one-off report
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
03
Official statistics
Governments, regulators and international organisations.
National Center for Education Statistics, US Department of Education
US federal statistics on degrees and certificates completed at every college taking part in federal student aid, by field, gender, race and residency.
Open dataUpdates: annual
6 charts
Public data; no licence stated. We use the figures as published in the AI Index Report 2026.
See the 6 charts
US Food and Drug Administration (FDA)
FDA's public list of AI-enabled medical devices authorised for marketing in the United States, with decision dates, specialties and companies.
Open dataChecked weeklyUpdates: periodic
5 charts
Public data; no licence stated. We use the figures as published in the AI Index Report 2026.
See the 5 charts
OECD education statistics (Education at a Glance, UOE data) on graduates by field and level, used for ICT graduates by country and gender.
Open data · CC BY 4.0Updates: annual
5 charts
Open data under CC BY 4.0. We use the figures as published in the AI Index Report 2026.
See the 5 charts
AI Action Summit, hosted by the French Presidency (Élysée)
Declaration from the February 2025 Paris AI Action Summit and its list of signatory countries and organisations.
Free to readUpdates: one-off
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
California Public Utilities Commission (CPUC)
Quarterly trip, mileage and incident data filed by companies in California's autonomous-vehicle passenger pilot and deployment programmes.
Open dataUpdates: quarterly
1 chart
Public data; no licence stated. We use the figures as published in the AI Index Report 2026.
See the 1 chart
International Energy Agency (IEA)
IEA special report on electricity demand from data centres and AI, and on AI's uses in the energy sector.
Free to readUpdates: one-off report
1 chart
Free to read; published under CC BY 4.0. We use the figures as published in the AI Index Report 2026.
See the 1 chart
OECD.AI Policy Observatory
OECD tool that detects and classifies AI incidents and hazards reported in global news media, using Event Registry news data.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
United Nations, High-level Advisory Body on Artificial Intelligence
UN advisory body's final report on global AI governance, including which countries take part in major international AI initiatives.
Free to readUpdates: one-off report
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
04
Company figures
Figures companies publish about their own products and users. They are not independent measurements.
Anthropic
Anthropic's privacy-preserving analysis of how people use Claude, by occupation, task and country, including automation versus augmentation.
Open dataChecked weeklyUpdates: irregular
2 charts
Public data; no licence stated. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Waymo's own comparison of crash rates for its driverless rides against human-driver benchmarks for the same miles and areas.
Free to readUpdates: periodic
2 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Baidu, Inc.
Baidu's investor-relations releases, which report fully driverless rides provided by its Apollo Go robotaxi service.
Free to readUpdates: quarterly
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Graphite (Graphite Growth)
SEO firm's analysis estimating the share of new English-language web articles written by AI versus humans since 2020.
Free to readUpdates: one-off study
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Microsoft AI Economy Institute
Microsoft estimate of the share of each country's working-age population using AI, built from anonymised Microsoft usage data.
Free to readUpdates: periodic
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
05
Benchmarks and leaderboards
Tests that score AI models. Scores depend on the test, the date and how each model was run, so each chart names both.
Company that runs its own evaluations of AI models on finance, legal, tax, coding and academic benchmarks.
Free to readUpdates: continuous
8 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 8 charts
Arena (formerly LMArena)
Crowd-sourced leaderboard where users vote between two anonymous chatbot answers; votes are turned into Elo-style ratings.
5 charts
Open data under CC BY 4.0 (leaderboard dataset). We use the figures as published in the AI Index Report 2026.
See the 5 charts
Artificial Analysis
Independent company that benchmarks AI models on intelligence, hallucination, price, speed and openness.
Free to readUpdates: continuous
2 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Video-MMMU team (Hu et al.)
Benchmark testing whether multimodal models can learn new knowledge from professional lecture videos.
Free to readUpdates: continuous
2 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Adaptyv Biosystems
Open protein-design competition in which AI-designed binders to a Nipah virus protein were made and lab-tested.
Free to readUpdates: one-off competition
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
ARC Prize Foundation
Puzzle benchmark of abstract visual reasoning that is easy for people and hard for AI, with cost per task.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Allen Institute for AI (Ai2)
Ai2's suite testing AI agents on scientific research tasks such as literature search, coding and data analysis.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Stanford Vision and Learning Lab
Robotics challenge scoring household-task success in simulation on the BEHAVIOR-1K benchmark.
Free to readUpdates: annual challenge
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
UC Berkeley, Gorilla project
Leaderboard testing how accurately models call tools and APIs, including multi-step and multi-turn calls.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Alek Safar and Oleg Chichigin
Benchmark testing whether AI models can read analogue clocks, a task trivial for people.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Stanford University (Zhang et al.)
Benchmark of professional capture-the-flag cybersecurity tasks for language-model agents.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Princeton HAL team; GAIA benchmark by Mialon et al. (Meta, Hugging Face)
Leaderboard for GAIA, a benchmark of real-world questions that require AI assistants to browse, use tools and reason.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Stanford CRFM (Center for Research on Foundation Models) with Arabic.AI
Stanford's Holistic Evaluation of Language Models extended to seven Arabic-language benchmarks.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Google DeepMind (Luong et al.)
Google DeepMind benchmark grading AI-written proofs to olympiad-level maths problems.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
AI4Bharat (IIT Madras)
Crowd-voting leaderboard comparing chatbots in Indian languages.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
MathArena (ETH Zurich SRI Lab; Balunović et al.)
Evaluates models on new maths competitions held after their training, to avoid contamination.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
OpenAI
OpenAI benchmark built from Kaggle competitions, testing AI agents on machine learning engineering work.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
TIGER-Lab (University of Waterloo)
Harder successor to the MMLU general-knowledge benchmark, with more answer options and more reasoning-heavy questions.
Free to readUpdates: irregular
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
MTEB project (Muennighoff et al.; community maintained)
Leaderboard ranking text-embedding models on retrieval, classification, clustering and other tasks.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
OpenGVLab (Shanghai AI Laboratory)
Benchmark of video-understanding tasks for multimodal models, such as recognising actions and their order.
Free to readUpdates: irregular
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
SloBench, Centre for Language Resources and Technologies (CJVT), University of Ljubljana
Leaderboard testing commonsense reasoning in standard Slovene and the Cerkno dialect.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
SWE-bench team (Princeton and Stanford researchers)
Benchmark testing whether AI systems can fix real GitHub issues in Python projects; 'Verified' is a human-checked subset.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Terminal-Bench team (tbench.ai)
Benchmark of realistic command-line tasks, such as compiling code or setting up servers, for AI agents.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Vchitect team (Zheng et al.)
Benchmark suite scoring video-generation models, with VBench-2.0 testing physics, common sense and faithfulness.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Leaderboard measuring how often models add unsupported facts when summarising documents, scored by Vectara's HHEM model.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Han et al.
Human-aligned benchmark for judging the quality of AI-generated video.
Free to readUpdates: irregular
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
WebArena team, Carnegie Mellon University (Zhou et al.)
Benchmark of realistic website tasks, such as shopping or forum admin, for autonomous web agents.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Sierra
Benchmark of AI agents talking to simulated customers and following company policy in retail, airline and telecom tasks.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
06
Research papers and studies
Academic papers and research groups. Each entry links to the paper or study.
RAISE Health (Stanford Medicine and Stanford HAI)
Joint Stanford Medicine and HAI initiative on responsible AI in biomedicine; it compiled the AI Index medicine chapter's literature and model counts.
Free to read
21 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 21 charts
Global Center on AI Governance (Adams et al.)
Expert-survey index scoring 138 countries on responsible-AI commitments, such as data protection, bias, gender equality and linguistic diversity.
Open data · CC BY 4.0Updates: irregular
5 charts
Open data under CC BY 4.0. We use the figures as published in the AI Index Report 2026.
See the 5 charts
Erik Brynjolfsson, Bharat Chandar, Ruyu Chen; Stanford Digital Economy Lab
Working paper using ADP payroll records to show falling employment for workers aged 22–25 in AI-exposed jobs such as software development.
Free to readUpdates: one-off paper
3 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 3 charts
Commentary estimating AI hardware power demand from chip-packaging supply-chain capacity, compared with countries, bitcoin mining and all data centres.
Free to readUpdates: one-off paper
3 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 3 charts
Augusto B. Corrêa, André G. Pereira, Jendrik Seipp
Paper comparing frontier language models with classical AI planners on PlanBench tasks, including deliberately disguised versions.
Free to readUpdates: one-off paper
2 charts
Free to read. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Study of who builds and downloads open AI models, using a Hugging Face download panel for the top models, 2022–2025.
Free to readUpdates: one-off paper
Rohit Saxena, Aryo Pradipta Gema, Pasquale Minervini (University of Edinburgh)
Benchmark paper testing whether multimodal AI models can read analogue clocks and answer calendar questions (ClockQA, CalendarQA).
Free to readUpdates: one-off paper
2 charts
Free to read; published under CC BY 4.0. We use the figures as published in the AI Index Report 2026.
See the 2 charts
Martina Francesca Ferracane, Simón González, Tomás Rogaler Wilson; European University Institute, Digital Trade Integration project
Dataset of 331 data localisation rules in 155 countries to December 2024, classed as local storage, local processing or conditional regimes.
Open data · CC BY 4.0Updates: one-off dataset
1 chart
Open data under CC BY 4.0. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Center for AI Safety, Scale AI and co-authors (Phan, Gatti, Li et al.)
Benchmark of expert-written questions across many academic fields, designed to stay hard for frontier AI models.
Free to readUpdates: continuous
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Paper estimating energy, water and carbon per prompt for dozens of language models from public API performance and hardware data.
Free to readUpdates: one-off paper
1 chart
Free to read; published under CC BY 4.0. We use the figures as published in the AI Index Report 2026.
See the 1 chart
The Project on Workforce at Harvard (Bick, Blandin and Deming, Real-Time Population Survey)
Tracker of US generative AI adoption from the Real-Time Population Survey, compared with the early spread of PCs and the internet.
Free to readUpdates: periodic
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Emma Strubell, Ananya Ganesh, Andrew McCallum
ACL 2019 paper estimating the carbon footprint of training NLP models, with everyday comparisons such as flights and car lifetimes.
Free to readUpdates: one-off paper
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Benchmark testing whether AI agents can answer research questions that need evidence from several papers and external tools.
Free to readUpdates: one-off paper
1 chart
Free to read; published under CC BY 4.0. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Chengzhong Ye, Gonzalo Benegas, Carlos Albors, Jianan Canal Li, Sebastian Prillo, Peter D. Fields, Brian Clarke, Yun S. Song
bioRxiv preprint introducing GPN-Star, a genomic language model, and comparing it with Enformer, Borzoi and Evo 2.
Free to readUpdates: one-off paper
1 chart
Free to read; published under CC BY-NC-ND 4.0. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Lei Zhang et al.
Benchmark of 23 multi-step robot tasks with electrical, fire and human hazards, scoring whether models finish tasks safely.
Free to readUpdates: one-off paper
1 chart
Free to read. We use the figures as published in the AI Index Report 2026.
See the 1 chart
Corrections
Spotted a wrong or missing credit?
Tell us and we will fix it, usually within a working day. Last reviewed 20 Oct 2026; the sources we rebuild from are checked every week.