OpenAI has introduced Astra for Law, a version of GPT-6 Astra built for legal work. According to OpenAI, the new service combines a specialized legal database with instructions for analyzing and drafting legal documents. The idea is to speed up some of the most repetitive tasks in law, from finding case law to preparing arguments, but it leaves us with more questions than answers.
A GPT designed for law firms
Astra for Law can search case law, statutes, regulations, court rules, and administrative decisions. It also integrates with tools like Relativity, Clio, or Thomson Reuters, for more context and a vetted information source. OpenAI wants AI to fit directly into the day-to-day work of law firms, with privacy controls for confidential information, but there’s more to consider about the launch than meets the eye.
First, this comes at a time when there are still several questions about the legal responsibility of AI agents, in short, who’s responsible when an AI system breaks the rules or even causes harm.
What could go wrong?
We’ve seen lawyers submit legal cases invented by ChatGPT, lawsuits caused by false information generated by AI, and, a few days ago, we saw a lawyer sanctioned for including fictitious witnesses in a murder appeal. The more convincing an answer is, the more important it is that we check every source and every line of reasoning.
Astra for Law can greatly improve legal research and help us find, compare, and draft faster, but the final decision must remain in human hands. Tools like these can change what technology is capable of, but, especially when we’re talking about rulings, evidence, or a person’s freedom, we can’t let it replace us, only help us.
