OpenAI launched Astra for Law on 17 September, moving beyond general-purpose AI into a product configured specifically for professional legal work.
The platform combines GPT‑6 Astra with specialised legal instructions and a search index covering US cases, legislation, regulations and court rules. OpenAI says the index draws on more than 230 million URLs and includes over 99.9% of published US precedential case law through the Free Law Project.
Initial access is limited to selected US firms. Harvey and Legora will be able to build products on the platform, while 26 integrations connect it with systems including Clio, iManage, Intapp, Relativity and Thomson Reuters. OpenAI also advertises confidentiality controls including zero data retention for eligible API users.
The strategic significance is larger than another legal chatbot. A foundation-model provider is now supplying legal research infrastructure while simultaneously powering specialist vendors and connecting directly to firms’ existing systems. That may compress the value of generic research and drafting features—and intensify competition over proprietary content, workflow integration and governance.
For Australian practices, immediate utility is constrained by the platform’s US-law focus. The broader direction is nevertheless clear: defensible legal AI will increasingly depend on jurisdiction-specific authority, verifiable citations, access controls and practitioner adoption of the reasoning—not model capability alone.
Sources: OpenAI announcement; Reuters.
Leave a Reply