OpenAI launches legal-focused AI platform, escalating race for law firm users

On September 17, 2026, OpenAI unveiled a specialized legal-focused iteration of its cutting-edge AI model, GPT-6 Astra, aimed at aiding law firms and legal technology providers in conducting research, drafting advice, and creating bespoke applications tailored for legal professionals. This launch signifies a burgeoning trend among leading AI companies entering the legal sector, where competition intensifies as tech firms vie for clientele among law firms and corporate legal departments.

In a recent development, Google broadened its Gemini Enterprise offerings tailored for legal practitioners, while Anthropic introduced tools designed for utilizing its Claude AI assistant since January.

OpenAI's latest offering integrates its Astra model with an extensive database encompassing US case law, statutes, regulations, and other pertinent legal resources, complemented by specialized guidance for legal analysis and documentation. The company further announced that legal AI startups like Harvey and Legora will have the capability to develop products leveraging Astra for Law, and the platform will feature integrations with leading software vendors such as Relativity, Clio, Intapp, and Thomson Reuters.

Thomson Reuters, the entity owning Reuters news service and the Westlaw legal research platform, debuted its own AI model, Thomson 1.0, last month. This model is trained on Thomson Reuters' legal research data and engineered for professional applications, built upon open-source technology and leveraging decades of the company's "authoritative" legal content.

Initially, a select group of law firms, including notable players such as Sullivan & Cromwell, Ropes & Gray, Cooley, Latham & Watkins, and Wachtell Lipton, will receive access to the platform under a special program, with rigorous measures to safeguard confidential client information. OpenAI disclosed that it has collaborated with these firms during the testing and development phases of legal AI applications.

The terms and conditions for utilizing the platform stipulate that all user comments must be in English, expressed in complete sentences, and devoid of abusive or personal remarks, adhering to the company's community guidelines.