Corporate Lawsuits Dataset 2019–2026

Key findings

1. 22,071 corporate lawsuits and enforcement cases involving 13,412 companies, January 2019 to June 2026. The company is the defendant in 89% of cases.
2. Case reports rose from 2,151 in 2019 to a peak of 4,260 in 2024. The largest types are civil (10,570), regulatory (5,593), class action (2,937), criminal (1,003).
3. A government body or regulator is the opponent in 2,862 cases (13%).
4. By jurisdiction: United States 10,321, India 2,900, United Kingdom 941, Canada 616, Germany 594 cases; the United States alone accounts for 50% of cases with a known jurisdiction.
5. The most-sued company is Google (76 cases as defendant), followed by Amazon (60) and Apple (56).
6. 2,494 cases report a settlement, fine or award, together $1.3T; the largest is BHP ($31.7B, 2024).

Lawsuits by case type and year

2026 covers January to June.

Top 50 largest settlements, fines and awards, 2019–2026

Amounts as reported, checked against the news sentence.

Top 50 largest claims and fines sought, 2019–2026

Most-sued companies (top 50)

Cases where the company is the first-named organisational defendant.

Lawsuits by jurisdiction country (top 40)

Lawsuits by country and year (top 30)

Lawsuits by sector (top 30)

Government bodies and regulators bringing the most cases (top 40)

Names as reported; one body can appear under several spellings.

Courts and tribunals hearing the most corporate cases (top 30)

Top 50 largest settlements and fines in the United States

Most-sued companies in the United States (top 50)

US lawsuits by sector (top 25)

Top 50 largest settlements and fines in the United Kingdom

Most-sued companies in the United Kingdom (top 50)

UK lawsuits by sector (top 25)

Top 50 largest settlements and fines in India

Most-sued companies in India (top 50)

India lawsuits by sector (top 25)

Column dictionary

23 columns in the sample CSV.

Method

1. Extraction. Radar's pipeline reads each news article, finds sentences about legal cases and extracts plaintiffs, defendants, case type, court, status and amounts. Upstream, reports carrying the same event fingerprint are grouped and retrospective mentions dropped.
2. Window and dates. The date is the publication date of the earliest report, read in Indian Standard Time; January 2019 to June 2026. Bulk archive back-fill days are dropped.
3. No people. A case is dropped when any plaintiff or defendant is a named person (recognised by name shape against a list of given names, or by a job title such as CEO or founder). Law firms announcing class actions are not treated as parties.
4. Company and opponent. The row's company is the first organisational defendant, else the plaintiff; the opponent is the first organisation on the other side. Opponents are typed as government / regulator from their name, else company; unnamed investors, consumers or employees are "shareholders / class". A website is kept only when the company record matches the reported name.
5. One row per case. Per company, in date order within a year: the same opponent, the same case type within 90 days, or sentences sharing half their words within 60 days are one case. The latest reported status is kept.

6. Amounts. Claims and settlements are converted to US$ and kept when they match a money figure in the sentence or are plausible with no figure to check; among reports, the value most others agree with wins. Values above $200B are blanked, above $20B they must be verified and reported more than once, and unchecked values above $1B are blanked.

7. Sample. Each company key is ranked by sha1(key). Keys are taken in rank order until the file holds max(25%, 5,000) rows (capped at 100,000).

8. Profile links. Links are built from Fliar's company index and checked against it. A row whose company has no public profile has an empty link.

Limitations

• Derived from news: unreported cases are missing; US securities class actions are heavily covered because law firms publish press releases about them.
• ClickHouse shards were partially unavailable at build time (shard 2 down; queries skip unavailable shards), so some cases may be missing. Article links resolve for roughly half of the rows because some article-store nodes were down.
• Excluding every case with a named person removes many disputes that also involve executives.
• Case type, status and party roles come from a classifier and some are wrong; one long case can be split into several rows when reports are more than a year apart.
• 2026 covers January to June only.

Cite this dataset

@misc{fliarbi_lawsuits_2019_2026_2026,
  author    = {{Fliar BI}},
  title     = {{Corporate Lawsuits Dataset 2019–2026}},
  year      = {2026},
  publisher = {Fliar BI},
  url       = {https://fliarbi.com/datasets/lawsuits-2019-2026/},
  note      = {Sample release. Licence CC BY-NC 4.0}
}