OpenAI Launches Astra for Law With 230 Million-URL Index
Legal work combines high-stakes judgment, vast case-law research and strict duties to protect client confidences, making citation quality and data governance key barriers to adopting general-purpose AI. OpenAI’s Astra for Law packages GPT-6 Astra with legal-specific instructions, a dedicated search index and connections to firms’ existing tools. The offering is designed as a foundation that law firms and legal-technology companies can adapt for research, contract review and transaction workflows, with confidentiality controls tailored to professional legal work.
OpenAI unveiled Astra for Law on Sept. 17, 2026, with an index spanning more than 230 million URLs covering U.S. case law, statutes, regulations, court rules and administrative decisions, updated daily. On 200 questions from a private Vals AI Legal Research Bench validation set, it passed the overall correctness check on 54.0% of questions, versus 38.7% for GPT-6 Astra using web search alone. Selected U.S. law firms will get initial access through Trusted Access in ChatGPT and Codex; an API is due later, with no launch date or price disclosed.
All Coverage
2 original reportsThe Backstory
The history behind this eventOpenAI’s Astra Becomes First Model to Hit Critical Cybersecurity Threshold
OpenAI’s Preparedness Framework tracks whether frontier systems develop capabilities that could create severe risks in areas including cybersecurity, biology and autonomous operation. Astra is the company’s first model to reach the framework’s “critical” cybersecurity threshold, a milestone that raises the stakes for access controls and deployment safeguards. The same capabilities that can help defenders find vulnerabilities and automate security work could also make sophisticated offensive techniques easier to use at scale.
OpenAI said its latest safety evaluation placed Astra at the critical cyber capability level, with reported testing showing the model clearing 48 CAPTCHA challenges in succession. The company has paired the release with tighter safeguards and a defense-oriented access strategy designed to preserve legitimate cybersecurity uses while limiting abuse. Astra’s launch marks the first time OpenAI has deployed a frontier model under protections calibrated for this level of hacking capability.
Subscribe to Mark Radar Weekly
Every Friday, the week's strongest signals in your inbox. Unsubscribe anytime.
If you search news on Google, you can set Mark Radar as a preferred source—our coverage will show up more often in your results. Set as preferred source on Google →