TL;DR
Big tech is now selling AI directly to law firms. OpenAI, Google, Anthropic, and Microsoft all have legal products. A federal appeals court ruled that training AI on Westlaw headnotes was not fair use. California also passed the first state law on lawyers’ use of AI. Meanwhile, courts keep punishing lawyers who file fake AI citations.
In June, lawyers on both sides of a Mississippi lawsuit cited cases that didn’t exist, and the judge canceled the trial. That story shows why AI legal tech news matters right now. AI legal tech is software that uses artificial intelligence for legal research, drafting, and document review. In recent weeks, OpenAI and Google launched legal AI products, California set new rules for lawyers, and a major court ruled on AI training.
What Is AI Legal Tech?
AI legal tech is software that uses artificial intelligence to help with legal work. Common jobs include legal research, drafting, contract review, and sorting documents in lawsuits. Some newer tools act as AI agents. An agent finishes several steps alone, such as finding cases, checking citations, and writing a first draft.
Two groups build these tools. Specialist companies, like Harvey and Legora, focus only on law. Meanwhile, big tech firms, like OpenAI and Google, now sell legal versions of their general AI. That second group drives most of this fall’s headlines.
What Is the Biggest AI Legal Tech News Right Now?
Four stories stand out as of October 6, 2026. Big tech launched legal AI products. A federal appeals court ruled on AI training and copyright. California passed the first state law of its kind for lawyers. And courts keep punishing fake AI citations.
Here’s a quick timeline:
| Date | News | Why it matters |
| Aug 25 | Google launches Gemini Enterprise for Legal | Big firms get a Google-built legal AI platform |
| Sept 17 | OpenAI launches Astra for Law | A top AI lab targets law firms directly |
| Sept 30 | 3rd Circuit rules for Thomson Reuters against ROSS | First federal appeals ruling on fair use in AI training |
| Sept 30 | Clio acquires Learned Hand | AI built for judges and courts is growing |
| Reported Oct 1 | California’s SB 574 signed | First state law on lawyers’ use of generative AI |
| Oct 2 | ROSS says it will seek Supreme Court review | The case could go higher |
How Are Big Tech Companies Entering Legal AI?
Big tech is now competing with specialist legal AI firms. Three launches show how fast this is moving.
What did OpenAI launch with Astra for Law?
On September 17, OpenAI launched Astra for Law. It isn’t a brand-new model. Instead, OpenAI wrapped its GPT-6 Astra model in a legal search index and legal writing instructions. The index covers more than 230 million web addresses of US case law, statutes, and court rules. Case law data comes from CourtListener, run by the nonprofit Free Law Project.
OpenAI tested the setup on 200 questions from a legal research benchmark by Vals AI. Astra for Law passed the overall correctness check on 54% of them. The same model using web search alone passed 38.7%. That’s a real gain. Yet it also means the tool missed nearly half of the questions, and OpenAI picked the benchmark itself.
Access is limited for now. Selected law firms can use it through a Trusted Access Program. An API version is coming later, with no date or price. Harvey and Legora can build on it, and plugins connect it to tools like Clio, Relativity, and Intapp.
What is Google’s Gemini Enterprise for Legal?
Google announced Gemini Enterprise for Legal on August 25. Cleary Gottlieb, Freshfields, Weil, and Williams & Connolly are launch customers. The platform handles drafting, citation checking, contract review, and tracking of regulatory changes. It also connects to LexisNexis, Thomson Reuters, iManage, and Everlaw. Harvey and Legora joined as launch partners.
Weil also announced BenchMark, an AI-enabled service built on Google’s platform. That detail shows a trend. Law firms don’t just use AI. They’re starting to package it as a service for clients.
Where do Anthropic, Microsoft, and Thomson Reuters fit?
Several others moved earlier. Anthropic released Claude for Legal in May, according to the ABA Journal. Microsoft added a Legal Agent to Word in April. Thomson Reuters, which owns Westlaw, launched an AI model called Thomson 1.0 in August. Reuters reports it’s trained on the company’s own legal content.
Why does this AI legal tech news matter for specialist companies?
Specialists now face giant rivals. However, many are partnering instead of fighting. Harvey and Legora joined both Google’s and OpenAI’s launches. Relativity’s CEO also said this week that the company will favor partnerships with frontier AI labs over legal AI startups. Therefore, the line between “big tech” and “legal tech” is getting blurry.
What Did the Court Decide in Thomson Reuters v. ROSS?
On September 30, the 3rd U.S. Circuit Court of Appeals ruled for Thomson Reuters against ROSS Intelligence. ROSS was an AI legal research company, and it has since shut down. The panel entered judgment affirming a 2025 summary judgment by Judge Stephanos Bibas.
That earlier judgment found two things:
- Westlaw headnotes were original enough for copyright protection.
- ROSS’s use of them to train its AI was not fair.
LawSites reports this is the first federal appeals decision to address fair use in AI training. However, the facts are narrow. ROSS copied the material before generative AI took off. It also built a non-generative tool that competed directly with Thomson Reuters. As a result, the ruling may not settle cases against OpenAI, Anthropic, or Meta.
The full opinion was initially sealed for redactions, so its reasoning could matter more once public. ROSS says it will ask the Supreme Court to review the decision.
What Are the New Rules for Lawyers Using AI?
What does California’s SB 574 require?
California’s SB 574 is the first state law written around how lawyers and arbitrators use generative AI. It passed the Senate 40-0 and the Assembly 75-0. Commentary published October 1 treats it as signed by Governor Newsom.
Under the law:
- Lawyers can’t delegate the practice of law to generative AI.
- Lawyers must keep confidential and personal information out of public generative AI tools.
- Lawyers must personally read and verify the authorities cited in papers they sign.
- Lawyers must disclose generative AI use to the court on documents they submit.
- Arbitrators can’t hand decision-making to generative AI.
Why do critics call the disclosure rule vague?
One attorney writing on the law notes it doesn’t say how to disclose, where to disclose, or what counts as “use.” He also points out that Google results now start with an AI summary. So does a quick search trigger disclosure? Nobody knows yet.
Other lawyers expect little change. One firm predicted the law won’t stop lawyers from using AI. Instead, it mainly reminds them to verify what they file.
Why Do Fake AI Citations Keep Reaching Courts?
AI tools can invent cases that sound real. This problem is called a hallucination. If a lawyer files without checking, the court may find the fake case first.
In June, a federal judge in the Northern District of Mississippi sanctioned four lawyers. Both sides had cited cases that didn’t exist. The judge canceled the trial. The two lawyers who drafted the filings were barred from the court for two years. They were also fined $2,500 and $3,500. One lawyer reportedly told the court she didn’t know AI could invent cases.
The Sixth Circuit also sanctioned two attorneys this year. Their briefs held over two dozen fake citations and misstatements. The ABA’s Litigation News noted that they also ignored an order to explain themselves, which made things worse. The lesson is simple. Verify every source, and answer honestly when a court asks questions.
How Fast Are Law Firms and Legal Departments Adopting AI?
Very fast. Law360 reports that an Association of Corporate Counsel (ACC) study found more than four in five legal departments have adopted AI. However, rules lag behind use. ACC’s report surveyed 576 in-house professionals and found these gaps:
- 41% don’t audit for unauthorized “shadow AI” use.
- 39% don’t formally track return on investment.
- 73% let non-legal departments, usually IT, give final approval.
What worries legal teams most?
Cost and accuracy top the list. DISCO’s annual study, released September 24, found teams are adopting AI under pressure from clients and leaders. Yet they fear usage-based billing because token and credit pricing is hard to forecast.
A separate Lexology PRO survey of more than 120 in-house teams found accuracy was the biggest worry. Only 14% had fully put AI governance frameworks in place.
Why are legal tech companies merging?
Bigger platforms are buying specialist teams to move faster. Clio acquired Learned Hand, which builds AI for judges and courts. Repario acquired UnitedLex. Opus 2 bought Bundledocs. Mitratech bought BotDojo to speed up its Aries AI software. In short, owning AI talent is now a race.
What Should Lawyers and Legal Teams Do Now?
This section is news analysis, not legal advice. With that in mind, the headlines point to a few practical steps:
- Verify every citation yourself. Open each case and read it. Never trust a summary alone.
- Check court rules. Many courts have standing orders on AI disclosure, and California now has a statute.
- Keep client data out of public AI tools. Use approved, private systems for confidential work.
- Write an AI policy and audit for shadow AI. Staff may already be using tools nobody approved.
- Ask vendors how pricing works. Get clear answers on tokens, credits, and usage caps before signing.
- Treat AI output as a draft. A human lawyer must own the final work.
What Comes Next for AI Legal Tech?
Four things deserve watching:
- ROSS’s Supreme Court request. If accepted, it could shape how courts treat AI training.
- OpenAI’s API release. Astra for Law is limited today, and wider access could change the market.
- How courts read SB 574. The disclosure rule needs clearer definitions.
- How pricing settles. Firms want predictable bills, and vendors are still testing models.
Conclusion
AI legal tech news is moving fast, and the pattern is clear. Big tech wants law firms as customers, courts are setting boundaries, and lawyers remain responsible for what they file. The tools are getting better. Even so, a 54% score on a research test shows why human checking still matters.
Pick one step from the checklist and finish it this week. Then bookmark the Techprixo AI section for updates, and share this guide with a colleague who works in law.
FAQ
AI legal tech is software that uses artificial intelligence to help with legal work. It supports research, drafting, contract review, and document sorting. Newer tools act as agents that complete multi-step tasks, like finding cases and checking citations. Lawyers must still verify the output before relying on it in court.
In recent weeks, OpenAI launched Astra for Law and Google launched Gemini Enterprise for Legal. California signed SB 574, which sets AI rules for lawyers. A federal appeals court also ruled for Thomson Reuters against ROSS Intelligence in a copyright case over AI training.
Yes, but they must check every citation themselves. Courts have sanctioned lawyers who filed fake AI-generated cases. California’s new SB 574 also requires lawyers to verify cited authorities and disclose to the court their use of generative AI. Rules differ by court, so always check local requirements first.
On September 30, 2026, the 3rd Circuit ruled for Thomson Reuters. It affirmed the lower court’s judgment, which found Westlaw headnotes protectable and ROSS’s AI training use not fair use. The full opinion was initially sealed for redactions, and ROSS plans to seek Supreme Court review.
Reliable sources include LawSites by Robert Ambrogi, Law360’s Legal Tech and AI section, Legaltech News from Law.com, and Artificial Lawyer. For rules and sanctions, also check state bar sites and court websites. Company announcements are useful, but compare them with independent reporting.