Updated
Updated · OpenAI · Sep 17
OpenAI Launches Astra for Law With 230 Million URLs and 40% Better Research Accuracy
Updated
Updated · OpenAI · Sep 17

OpenAI Launches Astra for Law With 230 Million URLs and 40% Better Research Accuracy

3 articles · Updated · OpenAI · Sep 17

Summary

  • Astra for Law combines GPT-6 Astra with a legal search index, legal-writing instructions and governance controls, and will first reach selected law firms through Trusted Access in ChatGPT and Codex before coming to the API.
  • On 200 U.S. legal research questions from Vals AI’s private validation set, the system passed overall correctness checks on 54.0% of questions versus 38.7% for GPT-6 Astra with web search alone—a 40% relative improvement.
  • The legal index searches more than 230 million URLs across U.S. case law, statutes, regulations, court rules and administrative decisions, with Free Law Project data covering more than 99.9% of published U.S. precedential case law.
  • OpenAI said eligible firms will get Zero Data Retention on the API and default exclusion from human review in ChatGPT Enterprise, while Latham & Watkins is helping shape permissions, ethical walls and oversight controls.
  • Harvey and Legora are among API customers building on the product, and OpenAI also launched 26 partner plugins plus 9 community plugins as it pushes a broader legal AI ecosystem.

Insights

With a significant failure rate on strict legal benchmarks, is Astra for Law truly ready to handle high-stakes corporate litigation?
Can zero data retention policies completely eliminate the risk of highly sensitive client data leaking into public AI models?