AI Regulation for Attorneys
Court rules, sanctions, and ethics guidance that govern how lawyers use AI.
Part of AI Regulation News, our running tracker of the laws, court rules, and agency guidance that change how professionals use AI at work.
For lawyers the AI regulation story has moved from theory to enforcement in a single year. The duty was always there. Rule 11 of the Federal Rules of Civil Procedure has long held that your signature on a filing certifies the law is real and stands for what you say. What changed is the price of skipping the verification step. Judges have stopped treating fabricated citations as an embarrassing slip and started treating them as sanctionable conduct that can cost money, a mandatory ethics course, and in the hardest orders the right to appear in a forum at all.
The pressure now arrives from four directions at once, and each one lands on a different part of your day. Court sanctions are the sharpest edge: a Mississippi judge revoked two lawyers' pro hac vice admissions and barred them from the district for two years over AI-invented cases, and appellate and state-court rulings are building a record that verification is non-delegable. Standing orders and local filing rules increasingly require an affirmative certification of how AI was used in a brief, which turns an informal habit into a documented obligation. Bar ethics guidance maps the older duties of competence, confidentiality, and supervision onto agentic tools, so the California State Bar's agentic AI guidance reads your duty under Rules 1.1, 1.6, and 5.3 onto autonomous workflows you may already be running. And the data-handling questions, from client confidentiality to whether a privileged prompt stays privileged, are being tested in real disputes rather than law-review hypotheticals.
Read through the desk of the lawyer who has to file on Monday, the common thread is verification discipline. Every one of these developments points at the same operational fact: AI changes the speed of drafting and changes nothing about who certifies the result. The competence duty now arguably includes understanding what a generative tool can and cannot do. The confidentiality duty governs what you are allowed to paste into a prompt and which tools meet your obligation to protect client information. The supervision duty means an autonomous agent gets treated like a non-lawyer assistant whose work you own. None of that is exotic. It is the existing rulebook applied to a faster pen.
What we track here is the part that actually alters your filing process, your engagement letters, and your firm's AI policy, and we skip the vendor announcements and academic debates that do not. Each entry below links to the primary order or opinion, states whether it binds you or merely signals where the trend is heading, and translates it into the one change it forces in how you work. If a development does not change a lawyer's duty, liability, or daily workflow, it does not earn a place on this page.
The regulation desk for attorneys
Each entry links to a full briefing with the primary source, the bindingness, and the one workflow change it forces. We add new entries as rules, rulings, and guidance land.
Delhi HC Refuses to Injunct OpenAI's Training on ANI Content
A single judge dismissed ANI's interim injunction application, holding on a prima facie view that storage for LLM training falls within section 52(1)(a) fair dealing. The court also found the outputs were not substantially similar and that memorisation was not proved. Paragraph 274 disclaims any bearing on the final outcome.
Electoral AIBrazil: an AI Avatar Does Not Escape a Speech Ban
An STF order applies the electoral deepfake rule, which bars synthetic audio or video of a real person used to harm or favour a candidacy even with that person's authorisation. Consent is not a defence where the rule protects a process.
Court SanctionsTenth Circuit Publishes Its AI Citation Rule and Declines Sanctions
A published Tenth Circuit opinion holds that fabricated citations breach Rule 46.5 and are sanctionable at the court's discretion, while declining to sanction in this case.
AntitrustThird Circuit Revives an AI Price-Fixing Case Against Casinos
A precedential Third Circuit ruling holds that competitors feeding non-public pricing and occupancy data into one vendor's AI engine can plausibly plead a Sherman Act section 1 conspiracy. It is a pleading standard, not a finding of liability.
Court SanctionsIllinois Sets $1,500 Per False AI Citation, and $15,000 in All
The First District fined an attorney $15,000 for 10 false citations, priced every type of falsehood at the same rate, and said fines must keep rising until they deter.
Court SanctionsMississippi Judge Bars Two Lawyers for Two Years Over AI Citations
Judge Aycock revoked pro hac vice admissions and imposed a two-year bar after both sides filed AI-fabricated cases, with a paste-ready pre-filing verification SOP.
Court SanctionsBrazil Court Sanctions Lawyer for AI-Fabricated Case Law
A Sao Paulo labor court (TRT-2) fined a party over AI-fabricated jurisprudence and referred counsel to the OAB-SP bar; the signer owns the citations.
Court SanctionsCalifornia Sanctions a Lawyer Over AI-Fabricated Citations
A published California appeals court fined a lawyer $1,500 and referred him to the State Bar for AI-fabricated case quotes, holding the duty to verify AI output is personal.
Court SanctionsNinth Circuit AI Sanctions: What Your Filings Now Require
How the appellate record on AI-fabricated authority raises the verification bar, and what an affirmative certification of AI use means for your briefs.
Court RuleFlorida's AI Filing Rule 2515 and the Sanctions Exposure
Florida moves toward an explicit rule on AI-assisted filings; what the proposed standing requirement asks you to certify and how it maps to existing duties.
Bar Ethics GuidanceThe California Bar Mapped Agentic AI to Your Supervision Duty
The State Bar's first agentic AI guidance read onto Rules 1.1, 1.6, and 5.3, with a supervision protocol for autonomous AI workflows.
Privilege LitigationThe Heppner Ruling and Whether Your AI Prompts Stay Privileged
A live dispute testing whether work product and privilege survive what you feed an AI tool, and how to structure prompts to protect the client.
Copyright SettlementThe 1.5 Billion Dollar Anthropic Settlement and the Fair Use Line
A federal judge drew the line between training on lawfully acquired books and hoarding pirated copies, and what it means for what you can safely feed AI.
StatuteEU AI Act Article 50: What Law Firms Owe by August 2026
The transparency and disclosure obligations under Article 50 that reach firms serving EU clients, with the August 2026 timeline and a compliance posture.
Preemption LitigationFederal vs State AI Preemption: Why You Still Comply Right Now
Why an executive order alone cannot preempt a state AI statute, the federal-versus-state status table, and a multistate compliance decision guide.
Browse the full AI Regulation News tracker for every entry across every profession, including topics outside this page.
Editorial note. This page curates our AI Regulation News coverage for attorneys. It is general information, not legal, tax, medical, or compliance advice. Each linked briefing carries its own primary sources, status, and last-checked date. Confirm against the underlying authority before relying on any entry.
Keep going
- Submission of the Office of the Privacy Commissioner of Canada to the Treasury Board of Canada Secretariat conNot binding. This is a regulator's submission to a Treasury Board Secretariat consultation on reforming the federal Privacy Act. It changes no law and creates no obligation. Any change would
- Legal services advisory AI Growth Lab: overview, published by the Department for Business, Innovation, ScienceNot binding, and not a safe harbour. This is a voluntary advisory sandbox. The overview states plainly that participation does not provide regulatory approval, endorsement or exemptions from
- Public consultation on the European Commission proposal for a Cloud and AI Development Act (CADA), Department Not binding on anyone. A consultation opened by the Department of Enterprise, Tourism and Employment to build an evidence base on a European Commission proposal that is itself still a propos
- Appointment of the Board of Oifig IS na hEireann (AI Office of Ireland) under the Regulation of Artificial IntAn appointment, not a rule. It creates no obligation on any provider or deployer, but it completes the governance layer of the body that coordinates Ireland's EU AI Act implementation.
- Principle Code on the protection of intellectual property and transparency for the appropriate use of generatiNot binding. The document states in terms that it is not a norm with legal binding force and does not compel disclosure of sensitive information. It operates by comply-or-explain, and it is
- Regulations on Personal Information Protection by Large Personal Information Processors (draft for comment), CNot binding. This is a draft circulated for public comment, and the effective-date clause in Article 50 is still blank. Nobody is obliged to do anything under it yet.
- Interim results of the 2026 series of special actions on personal information protection, CAC, MIIT and MPSThis is a progress report, not an instrument. It creates no new obligation. What it does is tell you where three regulators have been looking during 2026 and where they say they are going ne
- Cybersecurity Standards Practice Guide: AI Browser Security Practice Guide (draft for comment, v1.1-202607), dNot binding, and the comment period is over. A TC260 practice guide is a technical document, not a national standard and not a regulation, and this text is still a draft for comment.
- Announcement of the Cyberspace Administration of China on the Release of the Eighteenth Batch of Deep SynthesiThis is a publication of filings already granted, not a new rule. It creates no obligation by itself. The underlying duty to file comes from Article 19 of the Provisions on the Administratio
- Advisory Guidelines on Use of Personal Data in Generative AI, and Response to Feedback on Public Consultation Advisory Guidelines. They are guidance on how the Commission reads the PDPA, not a new statute and not an amendment to one. The consultation is now closed and the final Guidelines are publis
- Public Consultation on Digital Infrastructure Bill, MDDI and IMDA, 1 July 2026, with the draft Bill at Annex AA consultation on a draft Bill. It binds nobody. MDDI and IMDA state plainly that the document and the draft Bill are released only for consultation and do not represent the final legislatio
- Permintaan Masukan Publik terhadap Konsep Laporan Tahunan Penyelenggaraan Bidang Kecerdasan Artifisial, BlockcThis is a public consultation on a draft form, not a regulation. The draft concept binds nobody. The underlying reporting obligation sits in PP 28/2025 and Permen Komdigi 15/2025, not in thi
- Notice on the Operation of the AI Product and Service Confirmation Procedure (MSIT Notice No. 2026-50, enactedBinding. This is an enacted administrative rule (gosi) of the Ministry of Science and ICT, effective 21 July 2026. It is procedural: it does not itself impose AI safety duties, it governs ho
- Personal Information System Innovation Task Force and public policy proposal intake, Personal Information ProtNot binding, and not a draft bill. This is an agenda setting exercise: a task force plus an open intake of proposals. Nothing in it changes the Personal Information Protection Act.
- Ley Num. 163-2026 (P. del S. 441), amending Article 2(c) of Ley Num. 139-2011, Ley del Derecho sobre la PropiaBinding and already in force. The statute states it takes effect immediately upon approval, and the legislature's own record gives 2 August 2026 as both the approval date and the effective d
- Decreto 214/2026, por el que se modifica el Codigo Penal del Estado de Yucatan, en materia de acecho y delitosBinding. This is an enacted amendment to the Yucatan penal code, not a bill or a consultation. Its transitory article provides that it enters into force the day after publication in the stat
- Despacho Decisorio no 3/2026/SFI, Processo no 00261.004804/2026-54, Agencia Nacional de Protecao de DadosBinding on the addressee. This is a preventive measure issued inside an open supervisory proceeding, not a final sanction and not guidance. Discord had ten business days to appeal to the Sup
- Request by the Secretaria Nacional de Direitos Digitais, Ministerio da Justica, to the Policia Federal and theNot binding, and nothing is blocked by it. This is a request from one arm of the executive to the federal police and to ANPD, each of which decides for itself whether and how to act.
- Acordo de cooperacao tecnica TSE-AGU sobre inteligencia artificial nas Eleicoes 2026Not binding. This is a technical cooperation accord between two institutions, and the document it commissions is a guia de boas praticas, an orientation guide. It creates no new electoral of
- Garante per la protezione dei dati personali, provvedimento n. 577 del 23 luglio 2026, doc. web 10281021Binding measure, in force. A prohibition on further processing under Article 58(2)(f) GDPR plus a formal warning under Article 58(2)(b). No fine was imposed.
- CNIL and CIANum note exploratoire, IA agentique et protection des donnees personnelles: equation a inconnues mNot binding. This is an exploratory note co-written by the CNIL and the CIANum. It states positions and identifies problems; it creates no obligation and announces no enforcement position.
- UODO, Listy pytan inicjalnych: Ochrona danych osobowych w kontekscie systemow sztucznej inteligencji, 6 AugustNot binding. UODO states the questions are not a binding interpretation of the provisions and do not determine GDPR compliance, and that the answers do not have to be presented to the superv
- Ministerstwo Cyfryzacji announcement, Piaskownice regulacyjne AI, 10 August 2026This is a ministry announcement describing a mechanism created by the AI systems act signed by the President. The announcement itself binds nobody; it explains what the statute sets up. No c
- O peri Rythmiseos Ilektronikon Epikoinonion kai Tachydromikon Ypiresion (Tropopoiitikos) Nomos tou 2026, draftA draft amending bill in open public consultation. It binds nobody. Comments close 16 September 2026 at 23:50 on the e-Consultation platform.
- CTU press release of 31 July 2026 on the Article 50 AI Act transparency rulesA press release, not an instrument. It creates no obligation of its own and binds nobody. The obligations it describes come from Article 50 of the AI Act, which the regulator says takes effe
- MIRRI SR press release of 13 August 2026 on the reworked draft Slovak act on artificial intelligenceNot law, and not binding on anyone. This is a ministry press release about a bill that has just been reworked and sent onward in the legislative process. It still has to clear the Legislativ
- Government announcement of 7 August 2026 on the launch of the mihivatal.gov.hu websiteAn announcement of a website, not an instrument. It binds nobody and changes no obligation. What it does is give members of the public a named government address for AI grievances.
- Draft Law on the Implementation of the European Union Artificial Intelligence Act, project No. XVP-1564Not law. A draft law that has only passed the presentation stage in the Seimas and returns to the plenary in the autumn session.
- TTJA guidance page Tehisintellektisusteemid (Artificial intelligence systems), Estonian Consumer Protection anGuidance on an authority's website. It binds nobody, creates no new obligation, and describes TTJA's supervisory role in the future tense throughout.
- Avukatlar Icin Yapay Zeka Kullanimi Tavsiye Rehberi and Yapay Zeka ve Avukatlik Calistayi Sonuc Raporu, TurkiyA recommendation guide and a workshop report from a professional body. Neither is a disciplinary rule, and the announcement describes them as recommendations shared with the profession and t
- Law of the Republic of Azerbaijan No. 387-VIIQD of 21 April 2026 amending the Criminal Code, the Criminal ProcBinding. This is an adopted law signed by the President, not a draft or a consultation. The e-qanun record carries the status Quvvededir, meaning in force.
- Elections (Means of Propaganda) (Compliance with the disclosure duty for election propaganda that is a deep faBinding, and already operative. The rules were set by the chairman of the Central Elections Committee under section 2A2 of the Elections (Means of Propaganda) Law, 5719-1959, and the committ
- FCC Public Notice DA 26-786, WC Docket No. 18-89, ET Docket No. 21-232, EA Docket No. 21-233Binding and effective now. The Public Notice updates the Covered List maintained under section 2 of the Secure and Trusted Communications Networks Act, and under existing rule 47 CFR 2.903(a
- Letter from Governor Greg Abbott to PUCT Chairman Thomas Gleeson and ERCOT President and CEO Pablo Vegas, AuguA gubernatorial directive to two named recipients, not a statute or a rule. It carries no docket number and no notice-and-comment process. Its operative force depends on what the PUC of Texa
- ERCOT's Requests for Good Cause Exceptions Relating to Batch Zero Deadlines and Status Update on Additional MaThis is a request, not an order. ERCOT asked the Public Utility Commission of Texas to grant good cause exceptions under 16 TAC 25.3(b); the filing includes a Draft Proposed Order the Commis
- Facial recognition technology: a guide to assessing the privacy risks (OAIC), updated 29 July 2026Updated regulatory guidance, not a rule change. The Privacy Act 1988 is unchanged. The guidance describes the OAIC's regulatory application of the law and now incorporates the Administrative
- SRA Warning Notice on the misuse of AI by solicitorsThe SRA says AI has no separate legal personality and that reliance on an AI output is not a suitable defence for false citations put before a court.
- Indiana Supreme Court Order 26S-MS-8, amending Trial Rule 11Since 1 August 2026 an Indiana signature certifies that any cited legal authority is authentic, and the rule expressly covers AI-assisted filings.
- Open the AI court rules topic tracker
- NYC Bar Formal Opinion 2026-2 on AI recordingThe City Bar ethics committee advises that the default should be not to record, and that consent be obtained from every participant on calls with non-clients.
- X.AI LLC v. Ellison (D. Minn.) TRO deniedA federal judge refused emergency relief on timing grounds, leaving Minnesota's nudification ban undisturbed pending a 19 August preliminary-injunction hearing.
- Elliott v. New York Bariatric Group (Conn. Super. Ct.) prompt injectionA judge sanctioned a filer for hiding white-on-white instructions telling any AI that read the filing to agree with him.
- APIG refers Google AI Overviews to the French competition regulatorA test of whether commitments given in 2022 reach an AI feature built later, relevant to any content-licensing practice.
- Leiske v. Kidd (Del. Ch.) GenAI rule to show causeA firm, not just the signing partner, had to file affidavits on its written GenAI policies and training.
- UKJT Legal Statement on liability for AI harmsEngland's expert statement treats failing to use AI as a possible breach of the standard of care.