AI × Law

Aderant has released a live integration connecting its iTimekeep software with Harvey's AI platform. The integration automatically converts work performed in Harvey—including research, drafting, and analysis—into detailed draft time entries with AI-generated narratives for lawyer review.

Debevoise & Plimpton has hired Davis Polk & Wardwell lawyers Mari Grace and Kendall Howell as counsel to support the launch of a proprietary AI risk software product. Grace focuses on AI issues in litigation and investigations, while Howell advises on AI governance and commercial transactions.

ArentFox Schiff has introduced FoxAI, a multi-tool AI framework designed to deploy industry-specific AI capabilities across practice areas including sports, real estate, and retail. The framework integrates proprietary tools, licensed third-party AI platforms, and a secure document environment to customize workflows and legal analysis for clients.

New data from Thomson Reuters reveals that 32% of law firm AI decision-makers would reject a job offer from an organization lacking professional-grade AI tools, compared to less than 10% of non-decision-makers. Senior partners and AI budget-holders report significantly higher confidence in AI's positive impact on their compensation and career trajectories than junior associates do.

California's SB 574 has advanced through the state legislature, proposing strict rules for AI use in the legal practice. The bill prohibits delegating the practice of law or judicial decision-making to AI, bans inputting confidential client data into public AI systems, and mandates that lawyers personally verify all citations filed in court.

The Solicitors Regulation Authority issued formal guidance warning UK law firms and solicitors about regulatory breaches linked to unchecked AI outputs. The regulator emphasized that lawyers remain personally accountable for AI hallucinations submitted to court and warned that entering client data into public AI tools can permanently waive legal privilege.

Major law firms including Paul Weiss, Quinn Emanuel, and Proskauer Rose have reportedly explored discussions with private equity groups and bankers regarding outside capital. Elite firms are increasingly considering alternative business structures and management services organisations (MSOs) to fund tech, AI infrastructure, and international expansion.

The Tennessee Court of Appeals upheld an eviction decision after finding a pro se appellant submitted briefs containing fabricated quotes and incorrect dates generated by AI. Judge Andy Bennett noted the document bore the "hallmarks of artificial intelligence" and warned that AI remains an unreliable alternative to proper legal research.

A Nevada federal judge ruled that absolute judicial immunity protects a state court judge from civil liability even if she allegedly relied entirely on artificial intelligence to issue a ruling without discretionary human thought. The court held that issuing rulings remains a standard judicial function, leaving dissatisfied litigants to seek remedies through appeals or disciplinary proceedings rather than federal lawsuits.

A court ruled in Bartz v. Anthropic that training AI models on copyrighted books is fair use, but storing pirated copies is straightforward infringement. The case settled for $1.5 billion (roughly $3,000 per work), establishing a precedent for data provenance liability.

The Tennessee Court of Appeals upheld a lower court's eviction ruling after discovering a pro se litigant used generative AI to create fabricated legal references and misquotes. Judge Andy Bennett noted in the decision that AI's lack of real intelligence prevents it from being a reliable substitute for thorough legal research.

Legal tech provider Intapp released its agentic AI platform Celeste into general availability. The system is designed to enforce strict law firm compliance, ethical walls, and material nonpublic information rules by restricting AI access based on specific user permissions and leaving full audit trails.

A Massachusetts Superior Court judge ruled in Shealy v. Seaside Investments that ChatGPT outputs are not protected under the work product doctrine if generated without an attorney's direction or involvement. The court emphasized that AI is a tool, not a person or qualified legal representative, and its outputs do not inherently reflect protected legal mental impressions.

A law school has introduced a largely permissive generative AI policy. Under the new rules, students are broadly allowed to use AI to study legal concepts, generate hypotheticals, summarize cases, and identify key issues.

Am Law 100 firm Goodwin Procter unveiled a custom AI-enhanced tool for venture capital financing, marking the first product in a planned $25 million annual tech investment. Built on Anthropic's Claude platform, the firm intends to expand the technology to litigation, M&A, and private equity practice areas.

Harvey and research lab Engram have open-sourced 'Calderwood & Harkness,' a synthetic law firm corpus consisting of 250 client matters and 100 million tokens designed to benchmark AI agents on institutional knowledge retrieval. Testing revealed that top frontier models like GPT-5.6-sol and Opus-4.8 satisfied only about half of grading criteria, struggling to comprehensively search large-scale unstructured firm data.

Swedish legal AI startup Legora is in early-stage discussions to raise new capital at a valuation exceeding $10 billion, just four months after hitting a $5.6 billion valuation. The platform, backed by Nvidia, plans to use the capital to fuel its continued U.S. expansion and compete directly with American legal tech players.

Data compiled by researcher Damien Charlotin shows U.S. courts caught over 1,000 AI-hallucinated legal filings over the past year, with incidents rising sharply from 525 in 2025 to 724 so far in 2026. While pro se litigants account for most offenses, courts are increasingly responding with struck filings, disciplinary referrals, case dismissals, and financial penalties reaching up to $110,000.

A Massachusetts Superior Court judge ruled in Shealy v. Seaside Investments that AI-generated output is not protected under the work product doctrine unless created at the direction of counsel. The court distinguished cases involving pro se litigants, holding that a represented party's independent use of ChatGPT through a non-attorney third party fails to shield the materials or prompts from discovery.

A survey of 528 in-house legal leaders reveals that 18% of legal departments have not begun adopting AI, with 56% citing a lack of regulatory guidance as the primary hurdle preventing exploration. Unclear tool appropriateness and missing evidence of ROI were also listed as major historical barriers, far outweighing budget constraints.