OpenAI launches Astra for Law, a legal AI foundation with a 230-million-document case law index

OpenAI has introduced Astra for Law, a specialized legal AI foundation built on GPT-6 Astra, giving law firms a dedicated legal search index, custom legal-analysis instructions, and firm-specific privacy controls for confidential client work.

What's new

OpenAI's announcement states: "Today, we're introducing Astra for Law: a new foundation for law firms and legal technology companies to build AI products and workflows around their expertise." The offering pairs GPT-6 Astra with a legal search index covering "U.S. case law, statutes, regulations, court rules, and administrative decisions across a corpus of more than 230 million URLs, with sources added daily." A partnership with the Free Law Project brings a collection "covering more than 99.9% of published U.S. precedential case law" into the system.

On a 200-question U.S. legal research benchmark, OpenAI reports: "Astra for Law passed the evaluation's overall correctness check on 54.0% of questions, compared with 38.7% for GPT‑6 Astra using web search alone – a 40% relative improvement." The company also says Astra for Law "found 24% more reference cases" on case-law-focused questions and "retrieved up to 54% more relevant passages from the correct court opinions."

Beyond search, "custom instructions for legal analysis and writing guide Astra for Law in applying that research to the client's facts, developing arguments or deal terms, and identifying weaknesses and uncertainty." The product is "initially offered to selected law firms through Trusted Access in ChatGPT and Codex" as GPT‑6 Astra Law, with an API version coming as gpt-6-astra-law. For confidentiality, OpenAI is offering "Zero Data Retention (ZDR) on our API, and usage of ChatGPT Enterprise is excluded from human review by default," and is working with Latham & Watkins "to design for information permissions, ethical walls, client instructions, and firm oversight."

OpenAI is also launching "26 partner-built plugins that help firms go deeper with the tools and knowledge they already use," including connections to Relativity and Clio, plus "9 community plugins from lawyers and legal engineers at LegalQuants, LECG, and Skills.law, with 47 custom skills." Separately, "ChatGPT for Word" is now generally available for in-tool proofreading and edit suggestions.

Context

Early users named in the announcement include Harvey, Legora, Sullivan & Cromwell, Ropes & Gray, and Cooley. Sullivan & Cromwell built "an agreement analyzer that brings the firm's negotiating playbooks and selected precedents into the review of a new deal." Ropes & Gray created "a deal diligence system around how its lawyers work through a data room and decide what matters to the deal." Cooley built "GO Public to bring its capital markets expertise into how companies prepare to go public, from drafting the IPO filing to identifying the risks that deserve management's attention." Wachtell, Lipton, Rosen & Katz is also named as a partner bringing "litigation and corporate expertise" into the collaboration.

This extends a pattern of vertical-specific foundations OpenAI has been building on top of its general-purpose frontier models throughout 2026, following similar sector plays in financial services and cybersecurity.

Why it matters

Legal research has been one of the clearest early commercial wagers in enterprise AI — firms bill by the hour on exactly the research and drafting tasks a strong retrieval-and-reasoning system can compress. A verified 230-million-document case law index with a measured 40% accuracy improvement over generic web search gives OpenAI a concrete, benchmarked answer to legal-AI skeptics who have pointed to hallucinated citations as the industry's biggest liability. Landing marquee firms like Sullivan & Cromwell, Latham & Watkins, and Wachtell as design partners also signals that OpenAI is trying to own the enterprise legal vertical directly rather than cede it to legal-specific vendors like Harvey — even as Harvey itself is listed as an early API customer building on top of the new foundation.