90 Minutes · 1.5 Ethics Hours

    OUR CLE FOR LAWYERS

    AI in the Courts: Sanctions, Candor, and Compliance

    Using the COUNSEL Framework

    45 Slides
    22 Cases Tracked
    ABA Opinion 512

    45-Slide Presentation Deck

    Scroll through the complete CLE curriculum

    Opening & Objectives

    Slides 1–3
    Slide 1

    AI in the Courts: Sanctions, Candor, and Compliance

    Using the COUNSEL Framework

    • 90 Minutes · 1.5 Ethics Hours
    • Developed by Attorney Matthew A. Mishak
    Slide 2

    What This Course Is (and Is Not)

    • Ethics-focused risk controls for litigation teams
    • Anchored to ABA Formal Opinion 512 (2024)
    • Not jurisdiction-specific legal advice
    • A compliance system — not a technology demo
    Slide 3

    Learning Objectives

    • 1. Spot the "hallucination risk pattern" that drives sanctions
    • 2. Apply ABA Formal Opinion 512's core duties to courtroom workflows
    • 3. Implement a pre-filing AI verification protocol
    • 4. Draft a client-facing "AI use notice" aligned with Model Rules
    • 5. Build a firm policy using COUNSEL that reduces sanctions risk
    • 6. Understand the "AI evidence" track and proposed Fed. R. Evid. 707

    The Sanctions Menu

    Slides 4–8
    Slide 4

    The Trend Line

    From Novelty Sanctions to Routine Enforcement

    • 2023: First shock cases
    • 2024: Sanctions spread + appellate posture
    • 2025: Multi-lawyer stacks
    • 2026: Role-based fines + pro hac vice revocation
    Slide 5

    The Sanctions Menu

    • Monetary fines ($1,000 – $10,000+)
    • Fee-shifting to opposing counsel
    • Striking briefs / motions
    • Mandatory CLE / AI training
    • Client notice / service of order
    • Bar referral / self-reporting
    • Pro hac vice consequences
    Slide 6

    The Non-Delegable Signature Rule

    "If you signed it, you own it"

    • Rule 11 / Rule 9011 + inherent authority
    • Courts reject "I didn't draft it" defense
    • Verification is the signer's personal duty
    • Local counsel is not a shield
    Slide 7

    The Three Failure Modes

    • (a) Fake case — citation does not exist
    • (b) Real case, fake quote — citation exists but language is fabricated
    • (c) Real case, wrong proposition — citation exists but doesn't say what you claim
    Slide 8

    The "Second Mistake"

    Why Sanctions Escalate

    • Post-inquiry rationalizations
    • Slow correction or refusal to withdraw
    • "Spin" instead of immediate errata
    • Courts punish doubling-down harder than the original error

    Sanctions Casebook

    Slides 9–24
    Slide 9

    Wave 1 (2023): The Case That Taught the Profession

    Court

    S.D.N.Y.

    Citation

    Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023)

    Sanction / Remedy

    Monetary sanction + corrective measures

    COUNSEL Fix

    S (Scrutiny), O (Oversight), L (Learning)

    Slide 10

    Wave 2 (2024): Sanctions Spread

    Court

    2d Cir. / E.D. Tex.

    Citation

    Park v. Kim, 91 F.4th 610 (2d Cir. 2024); Gauthier v. Goodyear, 2024 WL 4882651 (E.D. Tex.)

    Sanction / Remedy

    Referral + $2,000 + AI education

    COUNSEL Fix

    S, O, U, L

    Slide 11

    Wave 3 Early 2025: "It's Not Just One Lawyer"

    Court

    E.D. Cal. / D. Wyo.

    Citation

    United States v. Hayes, 2025 WL 235531; Wadsworth v. Walmart, 348 F.R.D. 489

    Sanction / Remedy

    Tiered sanctions across drafter/signers

    COUNSEL Fix

    O, S, U

    Slide 12

    Bunce v. Visual Tech. Innovations

    Court

    E.D. Pa.

    Citation

    No. 2:23-cv-01740-KNS, 2025 WL 662398 (Feb. 27, 2025)

    Sanction / Remedy

    $2,500 + AI/ethics CLE

    COUNSEL Fix

    U, S, L

    Slide 13

    Nguyen v. Savage Enterprises

    Court

    E.D. Ark.

    Citation

    No. 4:24-cv-00815-BSM, 2025 WL 679024 (Mar. 3, 2025)

    Sanction / Remedy

    $1,000 (Rule 11 failure)

    COUNSEL Fix

    S

    Slide 14

    Dehghani v. Castro

    Court

    D.N.M.

    Citation

    782 F. Supp. 3d 1051 (D.N.M. 2025)

    Sanction / Remedy

    Monetary + training/self-reporting

    COUNSEL Fix

    S, L, O

    Slide 15

    Mid Central v. Hoosiervac LLC

    Court

    S.D. Ind.

    Citation

    No. 2:24-cv-00326, 2025 WL 1511211 (May 28, 2025)

    Sanction / Remedy

    $6,000 (adopted/reduced); "verification is non-optional"

    COUNSEL Fix

    S, U, L

    Slide 16

    In re Martin

    Court

    Bankr. N.D. Ill.

    Citation

    No. 24 B 13368, 2025 WL 2017224 (July 18, 2025)

    Sanction / Remedy

    Rule 9011 monetary + mandatory in-person AI training

    COUNSEL Fix

    U, S, L

    Slide 17

    Johnson v. Dunn

    Court

    N.D. Ala.

    Citation

    No. 2:21-cv-01701-AMM (July 23, 2025)

    Sanction / Remedy

    Sanctions + disqualification/referral consequences

    COUNSEL Fix

    O, S

    Slide 18

    OTG N.Y. v. Ottogi American

    Court

    D.N.J.

    Citation

    No. 2:24-cv-07209, 2025 WL 2671460 (Sept. 18, 2025)

    Sanction / Remedy

    $3,000 + strike/withdrawal consequences

    COUNSEL Fix

    S

    Slide 19

    Noland v. Land of the Free

    Court

    Cal. App.

    Citation

    114 Cal. App. 5th 426 (2025)

    Sanction / Remedy

    $10,000 — fabricated quotes in appellate brief

    COUNSEL Fix

    S, U

    Slide 20

    Lacey v. State Farm General Ins.

    Court

    C.D. Cal.

    Citation

    2025 WL 1363069 (May 5, 2025)

    Sanction / Remedy

    Fee-shifting / cost-of-cleanup

    COUNSEL Fix

    S, O

    Slide 21

    Ramirez v. Humala

    Court

    E.D.N.Y.

    Citation

    2025 WL 1384161 (May 13, 2025)

    Sanction / Remedy

    "Staff used the tool" pattern

    COUNSEL Fix

    O, S

    Slide 22

    Benjamin v. Costco Wholesale

    Court

    E.D.N.Y.

    Citation

    779 F. Supp. 3d 341 (2025)

    Sanction / Remedy

    Client service requirement + deterrence

    COUNSEL Fix

    S, U, N

    Slide 23

    Gardner v. Combs & In re Kenney

    Court

    D.N.J. / La. 5th Cir.

    Citation

    Gardner (Dec. 15, 2025); Kenney (Oct. 23, 2025)

    Sanction / Remedy

    Payment plan + self-reporting; CLE mandate + bar referral

    COUNSEL Fix

    S, N, L

    Slide 24

    Wave 4 (2026): Multi-Lawyer Sanction Stack

    Court

    D. Kan.

    Citation

    Lexos Media v. Overstock.com, No. 22-2324-JAR (Feb. 2, 2026)

    Sanction / Remedy

    Public admonishments + fines by role + pro hac vice revocation + policy certification

    COUNSEL Fix

    O, S, L

    Opinion 512 + COUNSEL

    Slides 25–34
    Slide 25

    ABA Formal Opinion 512 in One Sentence

    AI is permissible, but lawyers must still meet baseline duties

    • Competence (Rule 1.1)
    • Confidentiality (Rule 1.6)
    • Communication (Rule 1.4)
    • Supervision (Rules 5.1/5.3)
    • Reasonable fees (Rule 1.5)
    Slide 26

    COUNSEL Framework Overview

    • C — Confidentiality
    • O — Oversight
    • U — Understanding
    • N — Notice
    • S — Scrutiny
    • E — Equity
    • L — Learning
    Slide 27

    C — Confidentiality

    Model Rule 1.6

    • Never paste client secrets into public/consumer GenAI
    • Approved tools list: green/yellow/red by data sensitivity
    • Prompt hygiene: use placeholders, keep translation tables in DMS
    • "Assume prompts persist unless proven otherwise"
    Slide 28

    O — Oversight

    Rules 5.1 / 5.3

    • No unsupervised AI for legal authority generation
    • Two-lawyer rule: drafter + independent verifier
    • Signature block protocol: personal certification
    • Staff may summarize, but not generate citations without attorney check
    Slide 29

    U — Understanding

    Rule 1.1, Tech Competence

    • Train on failure modes: hallucinated cases, Franken-citations, fake quotes
    • Teach RAG vs. raw chat distinction
    • Ban "citation generation" prompts without mandatory source retrieval
    • "Asking for citations" is a known hazard
    Slide 30

    N — Notice

    Rule 1.4

    • Client notice trigger test: cost, confidentiality risk, strategy, or work allocation?
    • Engagement letter AI addendum template
    • Court disclosure: only when required by order/rule or to avoid misleading
    Slide 31

    S — Scrutiny

    Rules 3.3/3.1 + Rule 11/9011

    • No case cited unless a human opens and confirms: exists, says what you claim, quote is verbatim, pincites match
    • Maintain a citation verification log
    • Red flags: citation returned only by AI; no docket footprint; quote not found in 60 seconds
    Slide 32

    E — Equity

    Rule 1.5 and Fairness

    • Do not bill "AI time" as attorney time
    • Disclose AI tool cost pass-throughs
    • Document human review for AI-assisted triage, settlement evaluation, risk scoring
    • Watch for biased proxies in AI decisioning
    Slide 33

    L — Learning

    Continuous Improvement

    • Quarterly training refresh: 10-minute new-case roundup
    • Post-incident review: every hallucination → policy update + training vignette
    • CLE mandates now appearing as sanctions themselves
    Slide 34

    COUNSEL as a Litigation Pipeline

    • Intake → Drafting → Verification → Filing → Post-Filing Monitoring
    • Each step mapped to COUNSEL controls
    • Build it into existing workflow — don't bolt it on

    Court Rules & AI Evidence

    Slides 35–40
    Slide 35

    Standing Orders & Disclosure

    • Check before filing: judge standing orders, local rules
    • Require verification of computer-generated content
    • Some courts mandate AI use disclosure
    Slide 36

    AI Disclosure in Practice

    Treat like ghostwriting or outsourced drafting

    • Disclose when required by rule/order
    • Disclose when necessary to avoid misleading court/client
    • Don't over-disclose — avoid "disclosure theatre"
    Slide 37

    AI as Evidence: The Deepfake Problem

    • Moving from "future risk" to active rulemaking
    • Machine-generated outputs require authentication
    • Metadata preservation is critical
    Slide 38

    Proposed Fed. R. Evid. 707

    Machine-Generated Evidence

    • Would apply Rule 702(a)-(d) reliability requirements
    • Targets machine outputs offered without an expert
    • Requires foundation for model training/validation
    Slide 39

    Litigation Practice Tips for AI Evidence

    • Authenticate all machine-generated content
    • Preserve metadata and generation logs
    • Consider expert foundations
    • Build cross-exam questions about model training/validation
    Slide 40

    The Two Tracks of AI Risk

    • Track 1: Filings accuracy & candor — sanctions for hallucinated authority
    • Track 2: Evidentiary reliability — deepfakes, machine outputs, authentication
    • Both tracks are accelerating simultaneously

    Workshop & Takeaways

    Slides 41–45
    Slide 41

    Workshop: Build Your AI Filing Checklist

    Using COUNSEL

    • Break into pairs
    • Use the Pre-Filing AI Verification Protocol
    • Assign roles: Drafter, Verifier, Signer, Backup
    Slide 42

    Hypothetical #1: Paralegal Research Assistant

    • A paralegal uses ChatGPT for case research
    • Three citations are hallucinated
    • Who gets sanctioned?
    • What do you do Monday morning?
    • → Tie to Ramirez / Wadsworth / Lexos patterns
    Slide 43

    Hypothetical #2: Urgent TRO Filing

    • 24-hour deadline, AI-drafted brief
    • Speed vs. verification tension
    • What must happen before e-filing?
    • → The verification protocol doesn't bend for urgency
    Slide 44

    The Five Non-Negotiables

    • 1. If it's cited, you (a human) must open it and confirm it.
    • 2. If you signed it, you own it — delegation is not a defense.
    • 3. Treat AI as a drafting assistant, never as authority.
    • 4. Protect confidentiality by default — tools, data, consent.
    • 5. Train, audit, and update — courts now order "policy + proof."
    Slide 45

    Implementation Plan + Resources

    One Week to Compliance

    • Day 1: Adopt the Pre-Filing AI Verification Protocol
    • Day 2: Publish approved-tools list (green/yellow/red)
    • Day 3: Add engagement letter AI addendum
    • Day 4: Run 30-min training with case spotlights
    • Day 5: Add AI incident response playbook + tabletop exercise

    Sanctions Case Tracker

    22 cases mapped to COUNSEL controls (2023–2026)

    DateCourtCaseWhat BrokeSanctionRemediationCOUNSEL Fix
    2023S.D.N.Y.Mata v. Avianca, Inc.Fabricated citations via ChatGPTMonetary + corrective measuresCorrective distributionS, O, L
    2023Colo. O.P.D.J.People v. CrabillAI-generated fictitious authorityDisciplinary postureBar consequencesU, S, L
    20242d Cir.Park v. KimNonexistent authority on appealReferral consequenceAppellate disciplineS, O
    2024E.D. Tex.Gauthier v. GoodyearFailure to verify AI output$2,000 + AI educationEducation requirementU, S, L
    2025E.D. Cal.United States v. HayesFictitious authority citedSanctions + corrective distributionBenchmark for later courtsS, U, O
    2025D. Wyo.Wadsworth v. WalmartAI drafting without verificationTiered fines (drafter/signers)Signature-block accountabilityO, S
    2025E.D. Pa.Bunce v. Visual Tech.ChatGPT use admitted$2,500 + CLE requirementAI/ethics CLEU, S, L
    2025E.D. Ark.Nguyen v. Savage Enters.AI-generated citations$1,000 (Rule 11)Rule 11 enforcementS
    2025D.N.M.Dehghani v. CastroUnverified AI outputMonetary + trainingSelf-reporting elementsS, L, O
    2025C.D. Cal.Lacey v. State FarmRepeat AI failuresFee-shifting / cleanup costsSpecial master appointedS, O
    2025E.D.N.Y.Ramirez v. HumalaStaff used AI without oversightSanctions for supervising attorneyStaff supervision policyO, S
    2025E.D.N.Y.Benjamin v. CostcoUnverified AI citationsClient service + deterrenceClient notification requiredS, U, N
    2025S.D. Fla.Versant Funding v. TerasMulti-counsel AI failureInherent authority sanctionsMulti-counsel accountabilityO, S
    2025S.D. Ind.Mid Central v. HoosiervacRepeated fake citations$6,000 (adopted/reduced)"Verification is non-optional"S, U, L
    2025D. Colo.Coomer v. LindellAI-generated sanctionsReported sanctionsScrutiny enforcementS
    2025Bankr. N.D. Ill.In re MartinRule 9011 AI violationsMonetary + in-person AI trainingMandatory trainingU, S, L
    2025N.D. Ala.Johnson v. DunnUnverified AI outputDisqualification/referralNon-monetary consequencesO, S
    2025D.N.J.OTG N.Y. v. Ottogi Am.AI-generated false citations$3,000 + strike/withdrawalBrief strikingS
    2025La. 5th Cir.In re Kenney, KerryAI citations in appellate briefCLE mandate + bar referralAppellate disciplineS, L
    2025Cal. App.Noland v. Land of the FreeFabricated quotes/citations$10,000Appellate sanctionsS, U
    2025D.N.J.Gardner v. CombsStructured AI failuresPayment plan + self-reportingClient packet requirementS, N, L
    2026D. Kan.Lexos Media v. Overstock.comMulti-lawyer AI failuresFines by role + pro hac vice revocationPolicy certification requiredO, S, L

    Knowledge Check

    Q1. Under ABA Formal Opinion 512, which duty is most directly implicated by feeding client confidential information into a public GenAI tool?

    A: Confidentiality (Rule 1.6) and competence/supervision depending on safeguards.

    Q2. True/False: If your paralegal used an AI tool to generate citations, you can avoid Rule 11 exposure if you didn't know and didn't prompt the tool.

    A: False. Courts repeatedly treat verification as non-delegable for signing attorneys.

    Q3. Which COUNSEL element is the direct "antidote" to fabricated quotations?

    A: S — Scrutiny (verify quotes against the underlying source).

    Q4. Name two common remedial measures, besides money, that appear in hallucination-sanctions orders.

    A: Mandatory CLE/training; client notice/service; bar referral; striking briefs; pro hac vice consequences.

    Q5. Proposed Fed. R. Evid. 707 would push which reliability framework onto certain machine-generated evidence?

    A: The Rule 702(a)–(d) reliability requirements.

    Q6. Is it permissible to bill the client for time the AI tool spent generating a draft while you did other work?

    A: Bill only for your actual time and ensure overall reasonableness/transparency; not for fictional "AI time."

    AI Strategy and Governance for Modern Law Firms

    Founded by Matthew A. Mishak, Esq. — Harvard Business School Executive Education Graduate, MIT Sloan Artificial Intelligence Graduate.

    Mission

    LegalTek.ai proves you don't have to choose between speed and care, scale and quality, efficiency and ethics. Dedicated to closing the justice gap through ethical AI adoption.

    The COUNSEL Framework for Ethical AI in Law

    Mapped to and operationalizing ABA Formal Opinion 512 (the ABA does not endorse vendor frameworks). COUNSEL stands for: Confidentiality, Oversight, Understanding, Notification, Scrutiny, Equity, and Lifetime Learning.

    SilverTung AI Concierge

    A managed AI service for legal professionals with human oversight, ethical guardrails, and COUNSEL Framework compliance.

    Core Values

    Contact

    Website: https://legaltek.ai | Twitter: @legaltek_ai