Data accuracy disclaimer

Last updated 13 September 2026.

Please read this before relying on anything on the register. It applies to the free register, the verified records, the benchmarks and the reports.

1. Information only

The register is an information resource. It is not engineering, legal, insurance, investment or safety advice. It does not replace an incident investigation, a regulator's findings, an insurer's survey, or a professional's judgement. Decisions you take using the register are yours.

2. Where the information comes from

• Direct submissions come from people who were there or who worked on the asset. We review and anonymise them but we cannot independently verify every claim.
• Case studies are drawn from public sources: fire service reports, regulator filings, court records, operator statements, and trade press. Each cites its sources.
• Verified records (paid) are built from public sources and carry a status: unverified (one public claim, not yet corroborated), single source (one authoritative source such as a regulator or fire service), or verified (two or more independent sources agree on the key facts). The status is shown on every record.
• How entries are made. Most entries are found, drafted, reviewed and rechecked by an automated pipeline that uses language models, with an editor responsible for its rules and its output. The methodology page describes it in full. An entry labelled Provisional rests on thin sourcing and is rechecked automatically as more reports appear.

3. What the numbers can and cannot tell you

1. The register records incidents that were reported publicly or submitted to us. It is not a complete list of every incident that has happened. Absence from the register does not mean an asset, technology, supplier or country has had no incidents.
2. The register publishes counts, not rates. It knows the capacity of the sites where an incident happened and not the installed fleet behind each technology, so it does not publish an incidents per gigawatt figure: dividing one by the other returns the reciprocal of average site size rather than a frequency. Where a figure depends on capacity, the number of entries recording one is shown beside it. Capacity is stated far more often in public reports for battery storage and offshore wind than for grid assets, so coverage differs by technology and a thinly covered technology is described less well here, not proved safer.
3. Loss estimates, downtime and dates come from the best available source and may be revised. Figures given as ranges are ranges for a reason.
4. Root causes are as stated by the sources. Where an investigation is still open, the cause is provisional and marked as such.
5. Naming a site, operator, OEM or integrator in a verified record records a public fact about an incident. It is not a judgement of fault, quality or liability.

4. Named entries and anonymised submissions

Entries written from public reporting name the site, the place and the companies as the reporting names them, and never an individual person. Naming is a statement of what was reported, not a finding of fault, quality or liability. Entries from first-hand submissions are anonymised on purpose: details are removed or generalised. Do not try to identify the site, operator or people behind an anonymised entry, and do not treat any entry as evidence against a named party without reading its sources.

5. Errors and corrections

We correct errors when we find them or when they are reported, and log them on the corrections page. If you are an operator, owner or supplier named in an entry or a verified record and believe it is wrong, email editorial@energyincidentregister.com with the reference: we acknowledge within 2 working days and decide within 10.

6. No warranty

The register is provided "as is". We do not warrant that it is accurate, complete, current or fit for any particular purpose. The terms of use set out the limits of our liability.