Litigation support — briefs, memos & deposition preparation
163 evidence items
AI that drafts legal briefs, memoranda, and supports deposition and testimony preparation with research and argument structuring. Includes precedent-based argument generation and witness preparation; distinct from legal research which finds materials rather than drafting advocacy documents.
Overview
Litigation support means AI that drafts briefs and memoranda and prepares lawyers and witnesses for depositions. It is advocacy work, as distinct from research, which finds the law. It matters because the time saved at firms that have adopted it is real and large, and the major legal publishers now sell general-release tooling for it. Yet it remains a leading-edge practice and steady, because fabricated citations and quotations keep reaching courts, even from firms that run layered verification. Judges and legislators are answering with sanctions, disclosure rules and verification duties that lawyers cannot delegate. The deciding question is whether an ordinary team can get reliable output without heavy bespoke governance. Until one visibly does, the gains belong to firms that have spent years building their own checks.
Current Landscape
Thomson Reuters' CoCounsel anchors vendor scale, with more than one million professionals across 107 countries. Its Brief Builder offers purpose-built motion drafting with a four-checkpoint review architecture. A Forrester study commissioned by Thomson Reuters reports 400% risk-adjusted ROI over three years for a composite firm, with payback under six months. Clark Hill adopted CoCounsel enterprise-wide across 800 attorneys under a firmwide governance framework.
Frontier-model vendors are now selling legal drafting directly. OpenAI released Astra for Law, a GPT-6 configuration with a dedicated legal search index. OpenAI demonstrated it drafting a motion to dismiss. OpenAI reports that it passed 54% of questions on Vals AI's Legal Research Bench, against 38.7% for GPT-6 Astra using web search alone; these figures are self-reported. It is available to API customers including Harvey and Legora. Am Law 200 firms get a separate Trusted Access programme with zero data retention.
Named firms report large time savings from litigation drafting tools. Pierson Ferdinand reports 100-150 hours saved per litigation matter with Harvey. Hengeler Mueller completed a firmwide Harvey rollout in six weeks and reached 96-97% adoption, according to a vendor-side case study. Wilson Sonsini became the first Am Law 100 firm on a commercial LexText litigation drafting licence rather than a pilot. Ballard Spahr rolled out the Syllo litigation agent firm-wide. The Straits Institute catalogues 32 documented litigation AI deployments across named organisations.
Deposition tools have become products in their own right. Litem, formerly Deposely, reports attorneys at more than 50 Am Law 200 firms, over 20,000 legal professionals and more than 10,000 depositions a month. It also claims up to 75% prep-time reduction, which has not been independently benchmarked. DepoSim, AltaClaro's deposition simulator, has been adopted for training at Littler and three other BigLaw firms.
Litem's finances are much smaller than its reach suggests. Its SEC Form D shows $6,173,982 sold to 27 investors to fund Litem Agent, which drafts motions, memos and deposition summaries with citations. Its 2025 revenue was under $250,000. Its $10M figure is a projected trajectory for the end of 2026 based on signed contracts, not recognised revenue.
Adoption is also reaching small plaintiff practices. Romano Law Group, a personal injury firm with 5–10 attorneys, reports that a deposition summary that once took close to a full day now takes under 8 minutes on Anytime AI. The firm estimates the tool saves one partner 2 to 5 hours a week. Eve, which builds a drafting agent for plaintiff firms, has published its design: one agent loop of up to 150 iterations for each complaint or motion, grounded in case files.
Practitioners are drawing lines around what these tools can do. Daralyn Durie, a trial partner at Morrison Foerster, writes that AI already prepares deposition outlines, timelines and mock-trial scripts, and can suggest real-time questions during a live deposition. She argues it cannot read a jury or judge when to abandon a cross-examination. Citation verification is spreading into training: Boston University School of Law adopted Clearbrief for student brief writing. Clearbrief also names the American Arbitration Association and Dykema as users.
Hallucinated filings keep rising despite sanctions. Reuters reports that AI hallucinations have surfaced in at least 1,395 US state and federal court cases. These include a New Mexico defence lawyer who filed fabricated police witness testimony in a murder appeal. Charlotin's database documented 574 cases in the first half of 2026, containing 3,609 hallucinated items. The same database lists 1,783 cases globally.
Commercial integrated tools do not remove the verification burden. In Quinteros v. Harbor Distributing, a trial judge found eight fabricated quotations in briefs even though counsel had used Lexis, Lexis Protégé and Lexis Citation Check. The California Supreme Court declined to depublish the opinion. Legal Stack benchmarking puts hallucination rates at 8-17% on legal research and 6-13% on contract review across leading platforms. A Ninth Circuit discussion cited 17%-33% error rates in Westlaw and Lexis legal AI tools.
Courts are raising penalties as a deliberate deterrent. In Scott v. Illinois Human Rights Commission, an Illinois appellate court imposed a $15,000 sanction and signalled that fines would keep rising. Judge P. Kevin Castel told Reuters that the need for sanctions with greater deterrent effect is growing. A Los Angeles lawyer representing State Farm was fined $999.99 for seven nonexistent citations.
Binding rules are replacing judge-by-judge orders. The Tenth Circuit proposed a rule on September 18 requiring certification that a human reviewed any filing prepared with generative AI. Public comments close on October 18, and the rule would take effect on January 1, 2027. California's SB 574 mandates disclosure, citation verification and non-delegation of legal judgement in filings. Legal Stack counts more than 140 federal standing orders, and there is still no unified nationwide rule.
International regimes are tightening at the same time. An Irish High Court practice direction warns lawyers to verify every word or face sanctions. The Qatar International Court and Dispute Resolution Centre (QICDRC) has issued its own AI practice direction. The Tokyo Bar Association has published guidelines on generative AI in legal practice. The UK Court of Appeal in R v FGD addressed the dangers of AI witness coaching. The EU has delayed its AI Act rules for courtroom AI systems.
Testimony preparation carries a new discovery risk. In Conservation Law Foundation v. Shell Oil, a magistrate judge in the District of Connecticut ordered disclosure of the AI prompts an expert used to produce her report. The court held that the parties' Rule 29 stipulation was too narrow to protect them. Litigators are now adding requests for AI logs to expert discovery. This exposes AI-assisted expert preparation to cross-examination.
Measurement and trust now limit adoption more than capability does. Supio's survey finds that trust, not cost, is blocking AI adoption in plaintiff law. Harbor research finds that law firms are investing heavily in AI but none can measure its business or client impact. Firms without mechanical citation checking, logged workflows or the staff to verify every authority face rising sanctions. They also face certification duties and disclosure exposure, so they stay on the sidelines.
Tier History
Evidence (163)
— A frontier-model vendor enters legal drafting directly. Astra for Law demonstrates motion-to-dismiss drafting and is generally available to API customers including Harvey and Legora. Its benchmark gains are self-reported.
— Vendor-side case study of Hengeler Mueller's firmwide Harvey rollout: 96-97% adoption within six weeks. Its deposition-outline savings may be blended with other firms' figures, so confidence is reduced.
— BU Law has built Clearbrief's deterministic citation verification into student brief writing. The piece names the American Arbitration Association and Dykema as users, which confirms the listed tool is active.
— Vendor-published small-firm case study: Romano Law Group cut deposition summaries from about a full day to under 8 minutes and uses the tool for deposition outlines and to role-play opposing counsel.
— In Conservation Law Foundation v. Shell Oil, a court ordered disclosure of an expert's AI prompts despite a Rule 29 stipulation, which opens a new discovery exposure for AI-assisted testimony preparation.
158 more · latest 2026-09-18 →
— Independent Reuters reporting: AI hallucinations have surfaced in at least 1,395 US cases, including fabricated witness testimony in a murder appeal. Judge Castel calls for stronger deterrent sanctions.
— The Tenth Circuit proposed a rule requiring certified human review of AI-assisted filings from 1 January 2027, after sanctions for fabricated citations in appellate briefs. A negative regulatory signal.
— A Morrison Foerster trial partner credits AI with deposition outlines, timelines and real-time deposition questioning, but sets limits at trial advocacy and live cross-examination judgement.
— Form D shows Litem raised $6.17M to move from deposition prep into drafting motions and memos. It also shows 2025 revenue under $250,000 and that the $10M figure is a projection, not ARR.
— Independent research database: 32 deployments of litigation AI across 24 countries, including Greenberg Traurig (3,200 lawyers), Grupo Financiero Inbursa, Nelson Mullins, BakerHostetler, Cleary Gottlieb, Freshfields with named CoCounsel Legal and Harvey adoptions.
— Thomson Reuters launches Brief Builder within CoCounsel Legal for five motion types with 4-checkpoint review (facts, arguments, legal reasoning with Westlaw citations, formatted brief); lawyers control at each stage; emphasizes 'Fiduciary-Grade AI™' with full citation visibility.
— Litem rebranding signals productization: 50+ Am Law 200 firms, 20,000+ professionals, 10,000+ depositions/month, 75% prep-time reduction, $10M ARR on track; deposition-specific deployment at scale with documented ROI.
— Australian 300+ partner firm deployed Harvey for deposition prep and document analysis: 7,000 pages analyzed in 5 hours vs 3 weeks, 100-150 hours saved per matter; governance and security framework included in production rollout.
— Illinois appellate court imposed $1,500 per false citation ($15K total) for brief with ten hallucinated citations; court explicitly states existing fines insufficient to deter AI use and warns continued escalation of penalties necessary—signals intensifying liability barrier.
— Charlotin's tracking database: 574 documented US court cases in first half 2026, 3,609 hallucinated items total; error taxonomy: 53% fabricated citations, 26% misrepresented cases, 20% false quotations; trend acceleration signals ongoing adoption barrier at scale.
— California SB 574 (passed 39-0, August 31): first statute directly regulating lawyer AI use—requires disclose AI assistance, verify every citation before filing, don't delegate practice to AI, maintain client confidentiality. Creates regulatory floor for adoption.
— Wilson Sonsini took paid commercial license (not pilot) for LexText litigation drafting covering pleadings, discovery, motions, briefs, signaling shift from experimentation to binding AI procurement for core litigation workflows by Am Law 100 firm.
— Trial judge stated this is 'worst example of misconduct' ever witnessed; brief cited real cases but fabricated quotations; despite counsel using Lexis, Lexis Protégé, and Lexis Citation Check, no substantive verification occurred; California Supreme Court denied depublication.
— Official bar guidance establishes two-stage citation verification framework (mechanical existence check plus substantive legal review); emphasizes verification must be embedded in workflows, not reliant on individual attentiveness, defining governance standard for litigation documents.
— Named law firms (Rupp Pfalzgraf, A&O Shearman) reduced litigation research time from 17–28 hours to 3–5.5 hours per matter; Forrester study projects 400% risk-adjusted ROI over three years with payback under six months.
— Greenberg Traurig (3,000+ attorneys across 50+ offices) completed firm-wide deployment of next-generation CoCounsel Legal; specific metric: brief drafting time reduced from 20–30 hours to 2–3 hours per matter.
— EU AI Act Annex III explicitly classifies litigation support AI as high-risk (equivalent to regulated medical devices); compliance deadline moved to December 2, 2027, establishing formal governance tier for European deployment.
— Independent tech analyst coverage of Thomson Reuters' August 20 CoCounsel Legal GA with new Westlaw Brief Builder for litigation briefs; agentic redesign built on Claude Agent SDK with 1M users across 107 countries and 500K+ enterprise customers.
— Comprehensive adoption metrics: 41% of law firms use generative AI (up from 28% in 2025); 59% of legal professionals use AI for drafting briefs/memos; Westlaw AI 34% hallucination rate, Lexis+ 17%; 1,598 verified U.S. court cases with AI-fabricated citations by August 2026.
— Rigorous empirical evidence of adoption asymmetry: only 11% of public defender offices deployed AI-assisted legal research or drafting tools versus 71% adoption in BigLaw firms, quantifying structural access gap in litigation support deployment.
— Critical governance barrier: zero of approximately 50 law firm C-suite leaders surveyed could demonstrate consistent methods to measure whether AI investments deliver business or client value, despite 41% annual software spending growth.
— Survey of 47 AmLaw 200 litigation leaders: only 11% report active integrated AI deployment during live depositions; 60% report no real-time deposition AI use; identifies latency and architectural barriers.
— CoCounsel Legal benchmark on 50 realistic litigation tasks: completed in <8 minutes vs 6 hours for lawyers; attorneys rated output superior to expert answers on ~40% of tasks.
— Survey of 850 legal professionals: only 33% trust AI results despite 92% use; hallucinations cited as top challenge (48%); governance policies exist but only 30% applied effectively in practice.
— Legal ethics analysis: unverified AI use violates Rules 1.1 (competence), 8.4 (candor to court), and confidentiality duties; establishes compliance framework and professional liability exposure for litigation support.
— BigLaw analysis identifies three persistent AI limitations in litigation: context blindness, anchoring bias and model drift, and irreplaceable human judgment in witness assessment and investigative conclusions.
— Charlotin database: 1,868 global cases with documented AI hallucinations (1,297 US); 724 cases in 2026 YTD; demonstrates massive adoption scale with systematic failure pattern in litigation documents.
— Delaware Court of Chancery addresses briefs with fictitious citations and fabricated legal propositions; establishes Rule 11(b) violation standard for unverified AI output and non-delegable verification duty.
— Am Law 100 firm (750 lawyers) deploys agentic litigation platform firm-wide across all practice groups and case stages; offers fixed-fee options and consistency improvements enabled by AI integration.
— Irish High Court Practice Direction (effective Sept 1, 2026) mandates AI verification, disclosure, and non-delegable attorney responsibility for briefs, memos, and witness statements.
— QFC court comprehensive 27-paragraph AI governance framework for litigation: prohibits AI-generated evidence, requires citation verification against primary sources, establishes disclosure and penalties.
— England & Wales appellate ruling prohibiting AI-generated witness preparation materials; signals judicial restriction on AI-assisted deposition preparation and sets precedent for witness coaching liability.
— LexisNexis launching Protégé Vault for multimedia deposition analysis; lab achieving production-ready deposition capabilities in weeks rather than months, collapsing iteration cycles.
— Next-generation CoCounsel Legal (GA June 22, 2026) deployed across 1M+ users with partner feedback documenting capability improvements in complex legal analysis and briefing.
— Illinois appellate court imposed higher-than-typical fine for AI-hallucinated citations in litigation brief, signaling enforcement escalation and documentation of hallucination barriers to adoption.
— Sullivan & Cromwell hallucination filing analysis: two-layer safeguard (policy + review) failed simultaneously due to human attention under deadline pressure; mechanical verification layer needed.
— Independent benchmarking of hallucination rates across major legal AI platforms: 8–17% on legal research, 6–13% on contract review; documents accuracy risks for litigation support deployment.
— Regulatory analysis: 1000+ hallucination cases globally, California moving to enforceable rules, multi-circuit consensus on verification duty, and failure modes (suspensions, disqualifications).
— Independent legal tech analysis by Maestro Strategic Partners profiling CoCounsel's agentic redesign, core litigation-support capabilities (deposition prep, deep research), and deployment scope.
— Thomson Reuters reports 1M CoCounsel users across 107 countries; cites Forrester 2026 study showing 400% ROI for law firms; documents user-research-driven agentic redesign involving 500+ legal professionals.
— Survey of 207 plaintiff law firm leaders finds 30% regular AI use; trust in output (not cost or complexity) is primary adoption barrier; 99% reject unverifiable AI content.
— Comprehensive tracker of court AI disclosure and verification obligations mapping statewide policies, court-level rules, and 300+ individual judge standing orders; no nationwide rule exists.
— International law firm Clark Hill (800+ attorneys) announces enterprise-wide CoCounsel Legal deployment after 2+ years of governance development; emphasizes responsible AI adoption and ethics infrastructure as competitive advantage in enterprise litigation market.
— Comprehensive live-updated tracking database of 1,783 AI hallucination cases globally (1,228 US), structured by jurisdiction and outcome; serves as authoritative reference cited by courts, media, and researchers; updated through July 21, 2026.
— Comprehensive judicial enforcement response mapping 140+ federal district standing orders and 85+ documented sanctions cases for AI hallucinations in litigation filings through June 2026; establishes scale of institutional response to adoption barrier.
— Analysis documents stratified AI-driven deposition prep deployment across firm types; plaintiffs' contingency firms routinely deploy CoCounsel, Relativity aiR, Deposiq, EvenUp for witness history cross-referencing; defense teams lag due to hourly billing model, creating competitive advantage in deposition preparation.
— Thomson Reuters launches Deep Research Verify, automated citation verification system for legal briefs that validates cited authorities and highlights supporting passages; directly addresses hallucination/citation-error adoption barrier documented in practice evidence.
— AltaClaro CEO reports DepoSim adoption by four BigLaw firms (Orrick, Taft, Littler, Brownstein Hyatt) with specific outcomes: associates gaining experiential reps, senior lawyers maintaining skills between matters, partners deepening mentoring relationships; demonstrates production-scale adoption of specialized deposition training tool.
— CoCounsel Legal undergoes major agentic redesign on Claude Agent SDK with 100+ beta participants; positive feedback from law firms (Sterne Kessler, Troutman Pepper, Polsinelli); full GA planned August 2026, signaling platform maturity and vendor response to adoption barriers.
— Forrester validated study across 6 law firms documents 400% risk-adjusted ROI over 3 years, 25% matter capacity increase without headcount, payback under 6 months; demonstrates sustained production deployment value for litigation research and drafting workflows.
— Product feature releases targeting litigation workflows: Litigation Document Analyzer now generates counterarguments with supporting case law; Internal Investigation workflow for structured investigation reports; demonstrates vendor platform capability advancement specifically for litigation support.
— Survey of 500+ corporate legal leaders: 96% adoption breadth but only 31% at scale deployment; two-thirds stuck in pilots; major barriers include 49% citing provider fragmentation, 44% data security, 43% accuracy/reliability, 35% lack of expertise; reveals structural deployment ceiling preventing mainstream adoption.
— Independent Forrester economic impact study: large firm composite (500 attorneys) saw 31% matter volume increase, 82% reported time savings, 76% improved research/drafting quality, 53% improved client retention; demonstrates sustained production ROI for enterprise deployment.
— Practitioner analysis identifies data governance and centralization as prerequisites for AI effectiveness in litigation; most firms' data fragmented across drives/emails/systems creating unreliable AI input; documents implementation barrier that prevents effective adoption even when tools deployed.
— India's Supreme Court AI Committee released comprehensive regulatory framework explicitly permitting legal research and citation verification with mandatory human oversight; absolute prohibitions on algorithmic adjudication; signals systemic institutional recognition of litigation support as regulated governance domain.
— Named large firm Gibson Dunn (Thomson Reuters SYNERGY award winner) deploying CoCounsel for deposition timeline extraction, 50-state complex analysis, and redirecting attorney time to strategic work; demonstrates operational integration at litigation-focused enterprise.
— Ninth Circuit appellate sanction case (LNU v. Blanche) citing peer-reviewed research (Magesh et al., Journal of Empirical Legal Studies 2025) showing legal-specific AI tools hallucinate 17% (Westlaw) to 33% (Lexis) on representative queries; sanctions $2,500 each, 6-month suspension; establishes benchmark error rates anchoring verification barrier.
— London International Disputes Week 2026 panel (arbitrators, technologists, practitioners) documents AI deployment failures and success factors: failures reflect user error not tech limitations; success requires sophisticated data grounding and structured workflows not chatbots; human oversight essential for production litigation support.
— Litigation boutique Quinn Emanuel (1,100 lawyers) deploys three high-leverage AI surfaces: conversational matter intake replacing 90-minute client work + 12 hours associate effort into structured fact chronology; AI document review for multi-terabyte productions; conversational deposition prep with adversarial witness simulation identifying weak spots and divergences from documents.
— D.N.J. court documents AI-generated fabricated quotations from real cases and citations to non-existent cases in opposition brief; explicitly questions certification process.
— Major 1,000+ attorney law firm (Littler) deployed DepoSim simulator firmwide for deposition preparation; demonstrates production-scale adoption of specialized deposition support tool.
— Legal research firm achieved 99.2% citation accuracy and 78% weekly adoption using hybrid retrieval RAG; demonstrates that technical solutions to hallucination barrier exist and can achieve production reliability.
— Lacey v. State Farm: $31,100 sanction for brief with fabricated citations; attorneys delegated to CoCounsel, Westlaw Precision Drafting, and Gemini without verification; errors persisted after court warning.
— Thomson Reuters case studies: Bassford Remele saves 6-8 hours daily, Justly Prudent doubled caseload without hiring, aggregate 240 hours annually per attorney (6 work weeks) and 5x ROI; demonstrates sustained productivity gains in production deployment.
— S.D.N.Y. magistrate sanctions attorney $2,500 for fabricated case citations in litigation memoranda; court explicitly rejected tool-blame excuse, holding that manual verification would instantly reveal hallucinations.
— N.D. California: supervising attorney sanctioned $1,001 for failing to verify CoCounsel-generated citations in discovery filing; court held that supervision duty applies even when citation error originates downstream.
— Comprehensive 2026 landscape analysis covering brief writing, hallucination benchmarks (17-34%), regulatory requirements (EU AI Act high-risk classification, Colorado Act, Illinois employment law), and ethical governance.
— Supreme Court of Puerto Rico imposed $1,000 fine—first monetary penalty in jurisdiction for AI-fabricated citations—with formal warning that repeat conduct could trigger suspension, establishing verification duty regardless of tool type.
— Salem attorney William Ghiorso faced largest single-attorney AI sanction ($109,700 aggregate) for brief with 15 fabricated cases and 9 invented quotations; includes technical analysis of root cause (next-token prediction without source verification) and proposed solutions (source-ID marking, deterministic validation, retrieval-grounded citation).
— U.S. Court of Appeals for Third Circuit sanctioned attorney for filing brief with AI-generated summaries of 8 DEA adjudications (7 containing errors, 1 non-existent case); establishes circuit-level disciplinary precedent for unverified AI drafting.
— Ethicore synthesis: 1,227 documented global hallucination cases; 79% of legal professionals now using AI; Stanford benchmark shows 17% error rates (Lexis+ AI) to 34% (Westlaw AI); sanctions escalating (90-day suspensions, $15K+ per attorney) but deterrence failing despite rising penalties.
— Multiple federal court sanctions documented: Chinedu Obi case (N.D. Ill.) $9,750 for 13 hallucinated cases; Fletcher v. Experian (5th Cir.) $2,500 for 16 fabricated quotations; expanding pattern of litigation-specific AI failures.
— Vendor guide documenting 300+ hallucination cases globally with technical architecture of citation validation engines (parsing, database verification, ML pattern recognition) and pre-filing verification workflow integration for litigation briefs.
— EU AI Act (August 2, 2026 enforcement) explicitly classifies litigation support tools as high-risk, mapping specific compliance requirements and penalties—establishing regulatory barrier to scaled adoption in European market.
— Attorney Greg Lake suspended indefinitely for filing divorce brief with 57/63 defective citations (90.5% error rate), including 4 fabricated cases and 20 hallucinated references, violating competence and candor-to-tribunal duties.
— Independent comparative assessment of AI tools specifically for brief writing, memo generation, and related drafting tasks, with empirical quality assessment and citation accuracy analysis—key evaluation criteria for litigation support adoption.
— Sixth Circuit removed attorney from case, denied compensation, and referred for discipline for CoCounsel-generated brief with real cases but fabricated quotations, demonstrating that enterprise vendor tools marketed to prevent hallucinations still require full citation verification.
— Register analysis via Ethics Reporter: Q1 2026 sanctions reached $145K (highest quarterly total), yet rate still increasing; paradox revealed—61.6% of federal judges use AI without disclosure/training while prosecutors face career sanctions for same errors—structural governance gap preventing responsible adoption.
— Sixth Circuit sanctioned attorney for filing principal and reply briefs with CoCounsel-generated fabricated quotations and mischaracterized holdings. Court imposed denial of compensation and disciplinary referral, establishing recent high-stakes failure precedent for unverified AI brief drafting.
— Am Law 100 firm Taft (350+ litigators) completed firmwide rollout of DepoSim after pilot completion, with senior attorney Lynn Rowe Larsen confirming realistic feedback and actionability—demonstrating production deployment of specialized deposition preparation AI.
— Ethics Reporter documents record $109,700 Oregon sanction (highest US AI penalty to date), multiple high-profile cases including Phoenix Suns and New Orleans city attorneys, and escalating institutional responses including departmental AI policies—institutional recognition of systemic crisis.
— SurePoint report: 63% mid-sized firm adoption, 487 US court hallucination instances in 2025 (10x 2024 total); 81% of firm leaders report internal reliability concerns despite adoption acceleration—quantifies rapid growth divergence between adoption and failure rates.
— Detailed court case showing AI hallucinations across five separate filings (complaint, Daubert motion, motions to strike, expert evidence, responses) with fabricated statutory text and dozen+ fake quotations from deposition transcripts—demonstrating deep integration failures in litigation document workflows.
— SDNY judge imposed FRCP 11 sanctions and default judgment after attorney repeatedly filed motions and briefs with false AI-generated citations despite court warnings, establishing default judgment standard for unverified AI use in litigation.
— BriefCatch launched RealityCheck verification tool (deterministic + AI analysis) to detect hallucinated citations and false quotations in litigation briefs—vendor ecosystem response to hallucination crisis with demonstrated capability to catch errors missed by courts.
— CoCounsel reached 1 million users across 107 countries by February 2026, confirming transition from pilots to production deployment across legal, tax, compliance domains; integrated across Westlaw, Practical Law, Checkpoint, and Microsoft 365.
— Thomson Reuters CoCounsel achieved production deployment at scale: 1 million users across 107 countries, confirming transition from pilot to mainstream enterprise deployment for litigation research and drafting.
— Analysis by Clio Solutions Champion identifies that post-pilot adoption depends on workflow integration, governance infrastructure, and data handling alignment rather than technical capabilities; firms entering 'Phase Two' where adoption accelerates through embedded integration, not isolated tools.
— LexisNexis released Lexis+ with Protégé GA in the US, an integrated AI workflow platform with 300+ pre-built workflows including litigation drafting (motions to dismiss, discovery, deposition documents) and Shepard's citation verification integration.
— DepoSim AI deposition simulator piloted with six major law firms (Orrick, K&L Gates, McDermott, Littler, Taft, Brownstein Hyatt) generated 160+ hours of testing; 97% of participants strongly agreed on value, 94% indicated reuse intent for deposition training and preparation.
— Industry report citing Thomson Reuters Institute data: 75%+ of large law firms deployed at least one production AI tool; litigation and eDiscovery are leaders; early adopters reduced document review costs 70-90%, cut weeks to days; identified barriers: small firm cost, defensibility concerns.
— Reviews judicial sanctions against attorneys for AI-hallucinated briefs in 2026 (Lacey v. State Farm $26,100+fees, Mid Cent. Health Fund $15,000 recommended); documents persistent verification barrier: courts declare attorneys must verify all AI outputs before filing.
— New York appellate court sanctioned defence lawyer $5,000 for submitting brief with 23 fabricated cases; demonstrates that hallucination risks and court penalties persist unchanged in 2026, constraining mainstream adoption scaling.
— Law firm interviews document 30-40% time reduction in writing/reviewing cases and discovery acceleration from 9 months to 2 weeks; however, hallucination rates improved to ~1 in 10 from earlier 10 in 10, signaling progress but persistent verification burden.
— Thomson Reuters expands CoCounsel Legal to UK with deep research on Westlaw and Practical Law, featuring tabular analysis for document review; signals sustained vendor investment in international platform scaling and ecosystem maturity.
— LexisNexis announces commercial preview of Protégé AI Workflows with hundreds of pre-built litigation tools (motions to dismiss, discovery, deposition documents) and custom workflow builder, advancing major vendor's agentic litigation support capabilities.
— ADGM court sanctioned law firm AED 282,508 for defence with fictitious cases due to inappropriate AI use; international precedent establishing liability risks and highlighting that hallucination barrier persists globally in early 2026.
— Dykema law firm partnership with Clearbrief integrates AI-powered legal drafting tools into Microsoft Word, demonstrating real deployment by named mid-size firm and streamlined workflows for litigation document creation.
— Major Canadian law firm analysis documents AI's transformative potential in litigation (data review, drafting) while highlighting persistent risks: accuracy concerns, hallucinated citations, professional responsibility exposure, and court willingness to sanction.
— K&L Gates practitioner analysis argues for multidimensional AI value measurement (cycle time, adoption velocity, quality metrics) beyond hours saved; emphasizes validation skills to catch hallucinations and strategic institutional use.
— LexisNexis launches Protégé General AI with 'Best Fit' mode supporting Claude Sonnet 4.5 and GPT-5.1, specialized agentic agents, and Shepard's Citation Agent for verification-assisted litigation workflows and document drafting.
— Thomson Reuters launches beta Deep Research on Practical Law for agentic legal research and deepens CoCounsel-HighQ integration (1M+ users), with $200M annual AI investment signaling sustained platform expansion for litigation support.
— ACC survey of 657 in-house professionals shows GenAI adoption doubled to 52% in 2025 (from 23% in 2024), with 91% citing efficiency gains in drafting/research; however, 60% report no noticeable savings yet, indicating ROI quantification gaps.
— Analysis of 66 court opinions sanctioning AI misuse and 232 local rules addressing AI use; courts 'growing less lenient' with penalties including fines and dismissals; establishes pervasive adoption barrier and liability risk in Q4 2025.
— Tillion.ai analysis: 95% of generative AI pilots fail to deliver measurable ROI; 38% of in-house legal teams actively using AI; identifies 'pilot purgatory' and governance gaps as scaling barriers limiting mainstream adoption.
— Bassford Remele real-world deployment documented saving 6-8 attorney hours daily using CoCounsel and Westlaw Precision; 240-hour annual projection per professional represents substantive productivity validation across litigation practice.
— California appeals court sanctioned attorney $10,000 for 21 AI-fabricated citations in briefs; novel issue: opposing counsel dinged for failing to detect fake citations, expanding professional duty scope in AI era.
— Independent analysis documenting CoCounsel deployment at 45+ large firms (45,000 lawyers, 6 Am Law 10), Fisher Phillips achieving 5-hour research reduced to 5 minutes for slip-and-fall deposition preparation.
— Independent analysis of Clearbrief adoption: Microsoft legal teams deployment, 4-5 hours saved per brief, Falcon Law achieving 20% cost reduction in appellate work, validating efficiency for litigation support.
— Westlaw study documented 22 cases with AI-fabricated citations between June 30 and August 1, 2025; 40% of legal professionals cite accuracy as top concern; reinforces persistent hallucination barrier limiting mainstream adoption.
— Tracks 129 global hallucination cases in court filings (91 US), with 59 lawyer-sourced; US fines averaged $4,713 (range $100-$31,100); case count accelerating (32 in May 2025 alone), documenting scale of adoption barrier and liability exposure.
— UK commercial law firm (60 staff) implements CoCounsel across litigation, employment, corporate, and real estate; within six months demonstrates efficiency gains and competitive advantage, signaling sustained international mid-market adoption.
— Details 2025 sanctions against K&L Gates ($31,100 for 9 false citations via CoCounsel/Westlaw), Latham & Watkins (Claude.ai citation error), Butler Snow (ChatGPT fake cases); establishes that hallucination risks and high legal liability persist despite vendor advancement in 2025.
— Case studies from three law firms (Parker Taylor, Drake, Milon) document CoCounsel deployment for litigation with specific time savings: 5-8 hours per deposition and getting opposition to summary judgment motions '80% of the way there,' validating efficiency in deposition prep.
— Legal ethics debate on real-time AI deposition tools (Verbit Legal Visor, Deposely); Fisher Phillips deployed Legal Visor; ethical concerns raised about unauthorized practice of law and confidentiality, establishing adoption barriers in specialized deposition tools.
— Peer-reviewed Journal of Empirical Legal Studies study finds Lexis+ AI and Westlaw AI hallucinate 17-33% of the time, challenging vendor claims and validating core reliability barrier for litigation research and drafting tasks.
— Swedish law firm Morris Law announces firmwide adoption of CoCounsel as 'cornerstone' after 1.5-year evaluation, reporting streamlined workflows and increased precision for legal practice.
— Thomson Reuters launches next-gen CoCounsel to five new markets (Australia/NZ, Hong Kong, Japan, Southeast Asia, UAE) with 3x faster performance and deeper Westlaw/Practical Law integration, signaling sustained vendor commitment to scaling litigation support AI.
— Documentation of 2025 federal sanctions against attorneys for AI-hallucinated citations: Nguyen v. Wheeler ($1,000 sanction for fake cases), Bunce v. Visual Tech ($2,500 sanction plus ethics CLE), establishing that hallucination and verification barriers persist into Q1 2025.
— ABA survey of 512 attorneys shows AI adoption in law firms tripled to 30% in 2024; CoCounsel used by 26% of respondents, with civil litigation as a top adoption area (36% adoption rate).
— CLOC survey of 186 organizations shows AI adoption nearly doubled from 2023, with 30% of legal teams actively using AI and 54% planning adoption within two years in corporate legal departments.
— American Arbitration Association partnership with Clearbrief demonstrates real deployment with 20 arbitrators saving 8-10 hours per case on AI-assisted document drafting, providing independent-organization validation of efficiency gains in dispute resolution document work.
— Colorado appeals court decision on sanctions for AI hallucinations in court filings, extending sanctions precedent into Q4 2024 and reinforcing liability risks and verification burden as adoption barriers.
— Law firm counsel on hallucination risks in litigation, documenting recurring ethical failures and establishing that hallucination barrier persists unchanged in Q4 2024 despite vendor advancement.
— Coverage of Filevine's Depo CoPilot as real-time deposition analysis tool functioning as 'second chair,' demonstrating expanded adoption of AI-assisted deposition preparation in Q4 2024.
— Survey of legal professionals shows 76% of corporate legal departments and 68% of law firms use GenAI weekly for litigation work, with briefs/memos as top use cases, though barriers to scaling remain.
— Thomson Reuters launched AI for Justice Legal Aid program providing free CoCounsel access to select legal aid organizations and reduced pricing for nonprofits, expanding practitioner access and adoption.
— Thomson Reuters deepened CoCounsel 2.0 integration with Westlaw and Practical Law, adding Mischaracterization Identification in Quick Check and jurisdictional surveys to improve accuracy in litigation support.
— Filevine launches Depo CoPilot for real-time deposition transcription, goal tracking, and follow-up suggestions, creating specialized point solution for AI-assisted deposition preparation practice.
— Bar association ethics opinion establishes requirement for competency, verification, and disclosure in AI-drafted litigation documents, citing hallucination cases and emphasizing attorney accountability for accuracy.
— ILTA adoption survey reveals stark disparity in litigation support AI deployment between large firms (76% using in eDiscovery) and small firms (28%), indicating market segmentation and adoption barriers for smaller practitioners.
— CoCounsel 2.0 GA delivers 3x performance improvement with Claims Explorer for deposition prep and integration with Westlaw, Practical Law, and Microsoft 365, advancing major vendor's litigation support capabilities.
— LexisNexis adds Claude 3 and GPT-4o models plus GraphRAG integration with Shepard's citations for more authoritative litigation research and drafting, advancing competitor ecosystem for litigation support.
— Survey of 1,128 legal professionals shows 85% of law firms/in-house legal believe GenAI applicable to legal work, with briefs/memo drafting as top use cases, but only 12% of organizations using regularly.
— Thomson Reuters closes $650M acquisition of Casetext, confirming major vendor commitment to litigation support AI and consolidating CoCounsel as enterprise platform.
— U.S. Legal Support launches DepoSummary Pro, AI-generated deposition summary service that distills lengthy transcripts for case preparation, expanding specialist deployment of AI-assisted deposition support.
— Law firm analysis of GenAI litigation risks and mitigation strategies for companies deploying AI, cautioning that adoption creates liability exposure; emphasizes need for governance and contractual protections.
— Litigation practitioners report using GenAI for 'first pass on witness statements' to check consistency with documents, indicating active adoption of AI-assisted deposition preparation techniques.
— Thomson Reuters announces plans to deploy CoCounsel across entire product portfolio spanning legal, tax, risk and fraud; expanding toolkit availability across market segments.
— American Bar Association webinar demonstrating practical CoCounsel applications including Prepare for a Deposition skill, real-time examples of AI-assisted litigation document drafting and deposition prep.
— Judge imposed $2,000 sanction on attorney for submitting oppositions with AI-hallucinated case citations; emphasizes attorney obligation to verify AI-generated content before filing.
— CoCounsel Core expands to Canada and Australia after proven US deployment at 45+ law firms with 50,000+ lawyers and 9,000+ trained; includes specific deposition preparation skills.
— Survey data: 60% of attorneys used generative AI by mid-2023; memo drafting and document summarization identified as safe use cases; strategic use-case review recommended for adoption.
— Law firm analysis documenting hallucination risks and attorney-client privilege waiver dangers when using ChatGPT for legal work; recommends organizational AI use policies to mitigate adoption barriers.
— Stanford RegLab research finds hallucination rates of 69-88% on legal tasks; models fail 75%+ on complex reasoning like assessing precedential relationships, establishing fundamental reliability barriers to litigation support adoption.
— Deloitte analysis finds many legal teams stuck in proof-of-concept; identifies need for clear AI roadmaps, change management, and cost-benefit analysis as prerequisites for scaling adoption.
— Dubai International Financial Centre Courts issues official practice directive establishing AI governance in proceedings: transparency, accuracy verification, and attorney accountability.
— Another court filing with AI-fabricated case citations in table of authorities, documented as the 11th such incident in 2023, reinforcing critical reliability barrier to litigation adoption.
— Scholarly analysis in Judicature journal of judges' standing orders requiring AI disclosure and certification in filings, examining vagueness, overbreadth, and inconsistency as barriers to consistent AI governance.
— Ediscovery Innovation Report survey of 245 legal professionals: 12% actively using AI, 28% plan to use; 72% believe industry not ready; top concerns inaccuracy (44%), explainability (17%), security (14%).
— Clearbrief launches new AI features: Hyperlinked Timeline automatically generates case chronologies with source links, and Ask Your Docs queries document sets with verified answers, advancing deposition prep and brief support.
— AI legal writing platform Clearbrief integrates with legal practice management software MyCase, enabling automated brief writing and cite-checking within core law firm workflows.
— Law firm sanctioned $5,000 for submitting brief with six AI-fabricated case citations, documenting real-world failure and establishing precedent for AI liability in litigation.
— Law firm analysis documenting AI hallucination risks, judicial certification requirements, and adoption barriers following ChatGPT-generated fake citations in briefs.
— U.S. District Judge issues standing order requiring attorneys to certify AI use in filings or attest no AI was used, highlighting judicial response to hallucination risks.
— CLE webinar on AI technologies for virtual deposition management and preparation, signaling practitioner engagement and vendor activity in deposition support.
— Major law firm Fisher Phillips deployed Casetext CoCounsel firm-wide to 500+ attorneys for litigation research, document review, and drafting in production use.
— Clearbrief AI drafting platform integrates Fastcase legal research in Word for litigation document drafting, indicating ecosystem maturity and vendor collaboration.