AI labs pitch legal tools to lawyers as courts fine AI-made filings

Three of the four biggest AI labs have launched or pitched products for lawyers this year, even as courts keep fining attorneys over filings full of fabricated case law.

Legal products arrive faster than the fines

Google put Gemini Enterprise for Legal into preview in late August with four of the world's biggest firms, Gottlieb, Freshfields, Weil and Williams & Connolly. It came three months after Anthropic launched Claude Legal Solutions, which ships 20 connectors to legal platforms and 12 plugins and claims a 90.9 percent score on the BigLaw Bench reasoning test for Opus 4.7.

Elon Musk's SpaceXAI lab posted a legal page for its Grok models without disclosing how citations would be traced to primary sources. Cursor once pitched Grok 4.5 for legal work, but SpaceXAI's own announcement for the model discussed coding and agentic tasks instead. SpaceX bought Cursor for $60 billion on August 15.

Courts keep catching fabricated citations

The Ontario Law Society Tribunal found last December that lawyer Shahryar Mazaheri had used an earlier Grok version to draft a factum in a client's appeal, producing fabricated citations. A June costs decision ordered him to pay $31,150. Adjudicators said a language model "does not appreciate nuance or exercise judgment or use a moral compass" and called his materials "gibberish."

Sullivan & Cromwell, one of the highest-grossing US firms, filed an emergency letter with the Southern District of New York in April after a bankruptcy motion before Chief Judge Martin Glenn was found to contain 42 AI-generated fabrications. Researcher Damien Charlotin's database counts nearly 2,000 such cases worldwide. Canadian cases with fabricated citations rose from seven in 2024 to 86 in 2025, plus 39 in the first quarter of 2026.

Verification is still the lawyer's job

Google and Anthropic both present their products as built for verification. Google routes queries through connectors to external databases such as Everlaw and NetDocuments rather than relying on the model's training data alone. Google Cloud CEO Thomas Kurian said at launch that "ensuring every aspect of these agentic workflows is accurate, factual, and grounded in legal authority is of critical importance."

Independent evidence remains thin: a 2024 Stanford RegLab study found hallucination rates of 17 percent and 33 percent when testing legal AI from LexisNexis and Westlaw, and no comparable audit of Google's or Anthropic's tools exists. Bar rules put the duty to verify citations on the attorney who files them. Mazaheri's penalty came not from using AI, but from failing to check its output.