Open Australian Legal Corpus: 1.4 Billion Tokens of Legislative and Judicial Text

Isaacus released this corpus of 229,122 Australian legislative and judicial documents in 2026 for legal AI research. It contains over 60 million lines and 1.4 billion tokens sourced from the Commonwealth and various state-level legislative registers.

Use Cases

• Pre-training models for masked language modeling using the 1.4 billion tokens of legal prose
• Building document retrieval systems for Australian law using the 229,122 text records
• Generating legal summaries or drafting assistance based on the corpus of in-force statutes and regulations

Strengths

• 1.4 billion tokens of specialized legal text
• 229,122 individual documents
• Multijurisdictional coverage across seven Australian regions

Limitations

• No manual annotations for classification or entity recognition
• Excludes several major Australian jurisdictions like Victoria and ACT

Provenance

Source

Isaacus, with data from various Australian government legislative registers and courts

Collection Method

Found from official government sources

Time Range

Up to 2026

Freshness

Last updated March 2026.

Geography

Australia (Commonwealth and selected states/territories)