Federal law, federal courts and federal agencies, plus the ABA Model Rules as a national baseline. State professional-conduct rules, state courts and most private-law topics (family, property, estates, conveyancing, personal injury, construction liens) stay with the state where the matter is. US states are not modelled as separate jurisdictions on this site yet. Open the rules of the state of admission and the court on the file. Recorded 10 Oct 2026: Ethics Opinion 388: Attorneys' Use of Generative AI in Client Matters; Report and Recommendations of the Task Force on Artificial Intelligence; ABA news release on Formal Opinion 512. New York 22 NYCRR Part 161 (effective 1 June 2026) and the Unified Court System interim AI policy. New York City Bar Formal Opinions 2024-5 and 2026-2. Florida Bar Ethics Opinion 24-1 and Texas Professional Ethics Committee Opinion 705 stay the sources already cited on this page. No second source row was added. New Jersey preliminary AI guidelines (25 January 2024), Pennsylvania Joint Formal Opinion 2024-200 and the Illinois Supreme Court AI policy (effective 1 January 2025). Trackers of individual federal judges’ standing orders on generative AI. By September 2026 the pattern had shifted from judge-by-judge standing orders toward statewide rules, and reported AI hallucination cases had passed 2,000.
US lawyers are admitted and disciplined by a state (or the District of Columbia). Most states use a version of the ABA Model Rules. ABA Formal Opinion 512 (29 July 2024) is the national AI ethics baseline: Rules 1.1, 1.4, 1.5, 1.6, 3.1, 3.3, 5.1 and 5.3. State bars add guidance, for example California (2026), Florida 24-1, NYC Bar 2024-5, Texas 705, D.C. 388, Kentucky E-457, New Jersey and Pennsylvania 2024-200. The rules of your admitting state and your court bind you. Federal courts apply FRCP 11 to every signed filing. Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023) sanctioned fabricated ChatGPT citations. Some judges require an AI certificate. Proposed FRE 707 was not advanced on 7 May 2026. No general federal privacy law covers client data. Start with Rule 1.6, then HIPAA, GLBA or state law such as the CCPA (as amended by the CPRA). Colorado's SB24-205 was repealed and re-enacted by SB26-189 (signed 14 May 2026); ADMT rules start 1 January 2027.
Existing professional and judicial conduct rules apply fully; attorneys, judges and self-represented litigants are accountable for their work product; no disclosure mandate.
ARDC AI resources: https://ilcourtsaudio.blob.core.windows.net/antilles-resources/resources/e43964ab-8874-4b7a-be4e-63af019cb6f7/Illinois%20Supreme%20Court%20AI%20Policy.pdf
ABA news release on Formal Opinion 512Press releaseSecondary sourceUnverified dateOpinion 512 itself is already in the database; add only as a supporting source.Secondary. Guides barristers. Not a court practice note.
Formal Opinion 2024-5: Generative AI in the Practice of LawEthics opinionSecondary sourceUnverified dateThe linked page is Formal Opinion 2026-2, not 2024-5. Confirm 2024-5 on nycbar.org before reliance.Secondary. Guides barristers. Not a court practice note.
New York City Bar Association
2024 (per opinion number)
No confidential information into open AI systems without consent; verify outputs; do not bill hourly for time saved.
Ethics Opinion 24-1 (generative AI)Ethics opinionPrimary sourceUnverified dateChecked 10 October 2026: a source row already exists in the United States jurisdiction overview and four area-note source rows. This rule cites that source. The research file linked https://clearbrief.com/legal-ai-resources/florida, which is a secondary page and is not used as the source.Secondary. Scope not stated in the stored source.
The Florida Bar
approved by Board of Governors · Effective 2024-01-19
Treats AI like a non-lawyer assistant (Rule 4-5.3); fees must be reasonable (Rule 4-1.5); chatbots must identify as non-human.
Opinion 705 (generative AI)Ethics opinionPrimary sourceUnverified dateChecked 10 October 2026: a source row already exists in a United States area-note source row. This rule cites that source. The research file linked https://www.texasbarpractice.com/ethics-professional-responsibility/, which is a secondary page and is not used as the source.Secondary. Scope not stated in the stored source.
State Bar of Texas Professional Ethics Committee
issued · Effective 2025-02
Competence (Rule 1.01), confidentiality, supervision and independent verification of AI output.
Section 6068.1 was added to the Business and Professions Code by SB 574 (Stats. 2026, ch. 858). The Governor approved it on 30 September 2026, and it applies from 1 January 2027. It applies to California attorneys (lawyers) under the State Bar Act. An attorney shall not delegate the practice of law to generative artificial intelligence. An attorney who uses generative AI to assist in the practice of law must keep confidential, personal identifying and other nonpublic information out of systems where access is not restricted to the attorney and persons authorised by the attorney who are bound to protect it. The attorney must take reasonable steps to verify the accuracy of outputs, including all case and statutory citations, and to correct erroneous or hallucinated output. The attorney must also disclose the use of generative AI to the court for all documents submitted to the court. Not yet in force until 1 January 2027.
(a)(1) Nothing in this section shall be construed to abrogate an attorney’s duty to exercise reasonable competence and diligence in the practice of law.
(a)(2) An attorney shall not delegate the practice of law to generative artificial intelligence.
(a)(3)(A) Not enter confidential, personal identifying, and other nonpublic information into a generative artificial intelligence system for which access to confidential, personal identifying, or other nonpublic information the attorney inputs into the system is not restricted to the attorney and persons authorized by the attorney under obligations to protect the confidentiality of the information.
(a)(3)(B) Take reasonable steps to do both of the following: (i) Verify the accuracy of generative artificial intelligence outputs, including, but not limited to, the accuracy of all case and statutory citations. (ii) Correct any erroneous or hallucinated output in any material used by the attorney.
(a)(3)(C) Disclose the use of generative artificial intelligence to the court for all documents submitted to the court and consider whether to disclose the use of generative artificial intelligence if it is used to create content provided to the public.
(b)(2) “Personal identifying information” includes driver’s license numbers, dates of birth, social security numbers, National Crime Information and Criminal Identification and Information numbers, addresses and phone numbers of parties, victims, witnesses, and court personnel, medical or psychiatric information, financial information, account numbers, and any other content sealed by court order or deemed confidential by court rule or statute.
SB 574 amends Code of Civil Procedure section 128.7 so that a brief, pleading, motion or other paper filed in any California court must not contain citations that the responsible attorney has not personally verified. This includes citations provided by generative AI. It binds lawyers who submit filings. Breaches are subject to the section 128.7 sanctions framework, but monetary sanctions for a breach of paragraph (b)(2) may not be awarded against a represented party. Not yet in force until 1 January 2027.
(b)(2)(A) A brief, pleading, motion, or any other paper filed in any court shall not contain any citations that an attorney responsible for submitting the pleading has not personally verified, including any citation provided by generative artificial intelligence.
(c)(1) Absent exceptional circumstances, a law firm shall be held jointly responsible for violations committed by its partners, associates, and employees.
(d)(1) Monetary sanctions may not be awarded against a represented party for a violation of paragraph (2) of subdivision (b).
No new obligations; highlights RPCs: lawyers responsible for submissions, must not submit false content, must secure AI systems.
NJ Courts AI page: https://btlaw.com/en/insights/alerts/2024/new-jersey-judiciary-releases-preliminary-guidelines-for-unavoidable-use-of-ai-by-attorneys
This is the catalogue for United States. An instrument appears under “Rules that apply” only when you name the court or tribunal, or say that no court is involved.
Formal Opinion 2024-5: Generative AI in the Practice of LawEthics opinionSecondary sourceUnverified dateThe linked page is Formal Opinion 2026-2, not 2024-5. Confirm 2024-5 on nycbar.org before reliance.Secondary. Guides barristers. Not a court practice note.
New York City Bar Association
2024 (per opinion number)
No confidential information into open AI systems without consent; verify outputs; do not bill hourly for time saved.
Ethics Opinion 24-1 (generative AI)Ethics opinionPrimary sourceUnverified dateChecked 10 October 2026: a source row already exists in the United States jurisdiction overview and four area-note source rows. This rule cites that source. The research file linked https://clearbrief.com/legal-ai-resources/florida, which is a secondary page and is not used as the source.Secondary. Scope not stated in the stored source.
The Florida Bar
approved by Board of Governors · Effective 2024-01-19
Treats AI like a non-lawyer assistant (Rule 4-5.3); fees must be reasonable (Rule 4-1.5); chatbots must identify as non-human.
Opinion 705 (generative AI)Ethics opinionPrimary sourceUnverified dateChecked 10 October 2026: a source row already exists in a United States area-note source row. This rule cites that source. The research file linked https://www.texasbarpractice.com/ethics-professional-responsibility/, which is a secondary page and is not used as the source.Secondary. Scope not stated in the stored source.
State Bar of Texas Professional Ethics Committee
issued · Effective 2025-02
Competence (Rule 1.01), confidentiality, supervision and independent verification of AI output.
No new obligations; highlights RPCs: lawyers responsible for submissions, must not submit false content, must secure AI systems.
NJ Courts AI page: https://btlaw.com/en/insights/alerts/2024/new-jersey-judiciary-releases-preliminary-guidelines-for-unavoidable-use-of-ai-by-attorneys
Existing professional and judicial conduct rules apply fully; attorneys, judges and self-represented litigants are accountable for their work product; no disclosure mandate.
ARDC AI resources: https://ilcourtsaudio.blob.core.windows.net/antilles-resources/resources/e43964ab-8874-4b7a-be4e-63af019cb6f7/Illinois%20Supreme%20Court%20AI%20Policy.pdf
ABA news release on Formal Opinion 512Press releaseSecondary sourceUnverified dateOpinion 512 itself is already in the database; add only as a supporting source.Secondary. Guides barristers. Not a court practice note.
Section 6068.1 was added to the Business and Professions Code by SB 574 (Stats. 2026, ch. 858). The Governor approved it on 30 September 2026, and it applies from 1 January 2027. It applies to California attorneys (lawyers) under the State Bar Act. An attorney shall not delegate the practice of law to generative artificial intelligence. An attorney who uses generative AI to assist in the practice of law must keep confidential, personal identifying and other nonpublic information out of systems where access is not restricted to the attorney and persons authorised by the attorney who are bound to protect it. The attorney must take reasonable steps to verify the accuracy of outputs, including all case and statutory citations, and to correct erroneous or hallucinated output. The attorney must also disclose the use of generative AI to the court for all documents submitted to the court. Not yet in force until 1 January 2027.
(a)(1) Nothing in this section shall be construed to abrogate an attorney’s duty to exercise reasonable competence and diligence in the practice of law.
(a)(2) An attorney shall not delegate the practice of law to generative artificial intelligence.
(a)(3)(A) Not enter confidential, personal identifying, and other nonpublic information into a generative artificial intelligence system for which access to confidential, personal identifying, or other nonpublic information the attorney inputs into the system is not restricted to the attorney and persons authorized by the attorney under obligations to protect the confidentiality of the information.
(a)(3)(B) Take reasonable steps to do both of the following: (i) Verify the accuracy of generative artificial intelligence outputs, including, but not limited to, the accuracy of all case and statutory citations. (ii) Correct any erroneous or hallucinated output in any material used by the attorney.
(a)(3)(C) Disclose the use of generative artificial intelligence to the court for all documents submitted to the court and consider whether to disclose the use of generative artificial intelligence if it is used to create content provided to the public.
(b)(2) “Personal identifying information” includes driver’s license numbers, dates of birth, social security numbers, National Crime Information and Criminal Identification and Information numbers, addresses and phone numbers of parties, victims, witnesses, and court personnel, medical or psychiatric information, financial information, account numbers, and any other content sealed by court order or deemed confidential by court rule or statute.
SB 574 amends Code of Civil Procedure section 128.7 so that a brief, pleading, motion or other paper filed in any California court must not contain citations that the responsible attorney has not personally verified. This includes citations provided by generative AI. It binds lawyers who submit filings. Breaches are subject to the section 128.7 sanctions framework, but monetary sanctions for a breach of paragraph (b)(2) may not be awarded against a represented party. Not yet in force until 1 January 2027.
(b)(2)(A) A brief, pleading, motion, or any other paper filed in any court shall not contain any citations that an attorney responsible for submitting the pleading has not personally verified, including any citation provided by generative artificial intelligence.
(c)(1) Absent exceptional circumstances, a law firm shall be held jointly responsible for violations committed by its partners, associates, and employees.
(d)(1) Monetary sanctions may not be awarded against a represented party for a violation of paragraph (2) of subdivision (b).
Confirm the client’s instructions, jurisdiction, intended output and the information available before dividing the work.
Verify the authorities
Open each relied-on authority independently. Confirm its existence, current status, proposition and any quotation or pinpoint reference.
Check the filing requirements
Check the current court or regulator requirements, prescribed form and attachments for the intended document and filing.
Jurisdiction
Apply this guidance to the actual matter and jurisdiction. Confirm current requirements before relying on an example.
What to open first
Name the court or agency, then your state of admission.
Open the forum's local rules and any standing order on generative AI.
Open ABA Formal Opinion 512 and your admitting state's ethics opinion or practical guidance.
A New York City Bar opinion does not govern a Texas filing. Opinion 512 does not displace Rule 11 or a judge's certificate requirement.
Court and tribunal AI rules
FRCP 11(b): signing certifies legal contentions are warranted by existing law after reasonable inquiry.
Mata v. Avianca, Inc. treated fabricated ChatGPT citations as a Rule 11 problem: tools are not improper, but the lawyer is the gatekeeper.
Some judges add standing orders. E.D. Tex. (9 April 2025) requires a Certificate of Generative Artificial Intelligence Usage; without one, a filing is presumed AI-free.
EOIR PM 25-40 (8 August 2025): no ban and no mandatory disclosure, but judges may adopt standing orders. Hallucinated citations can support discipline under 8 C.F.R. § 1003.102.
Proposed FRE 707 was not sent forward on 7 May 2026. The draft was revised; deepfakes remain under study.
There is no single US court AI practice note.
Professional-conduct rules
State rules vary. Use Opinion 512 as the map, then read the local opinion.
California 2026 Practical Guidance (replacing November 2023): keep professional judgment; no agentic filing, advice or communication without review; no confidential input to risky tools without consent.
Florida 24-1: reasonable confidentiality steps (including whether the tool is self-learning), oversight policies, reasonable fees and verified research.
NYC Bar 2024-5, Texas 705 (February 2025), D.C. 388, Kentucky KBA E-457 (March 2024) and New Jersey's January 2024 preliminary guidelines cover the same ground locally.
Pennsylvania Joint Formal Opinion 2024-200, cited in Opinion 512, flags conflicts if a model reuses one client's information for another.
California has proposed AI-related amendments to its Rules of Professional Conduct. Check whether they are adopted before treating them as in force.
Privacy and confidentiality
Model Rule 1.6 (and the state equivalent): reasonable efforts to prevent unauthorised disclosure of information relating to the representation.
Opinion 512, Florida 24-1, California, Texas 705 and D.C. 388: public or self-learning tools are a confidentiality problem unless terms, retention and security are adequate or the client consents.
No general federal privacy law. HIPAA covers covered entities and business associates handling PHI.
CCPA, as amended by the CPRA: rights to know, delete, correct, opt out of sale or sharing, and limit use of sensitive personal information.
Colorado SB26-189 (signed 14 May 2026) repeals and re-enacts SB24-205; ADMT duties from 1 January 2027. HB26-1263 (signed 1 July 2026) adds chatbot duties from the same date.
Open the statute that covers the data, not a generic 'US privacy' label.
Review checklist
Open the court's or agency's current local rules, standing order or policy memorandum on generative AI.
Open Opinion 512 and the admitting state's opinion or guidance.
Verify every citation, quotation and record reference in an official reporter, court site or paid database, not another generative tool.
Read the output against the instructions and the source before filing. Rule 11 and Rule 3.3 are not delegated.
Keep client information, tax-return information and PHI out of public or self-learning tools unless the rules allow it.
Bill hourly work for time actually spent (Rule 1.5; Opinion 512, Florida 24-1, California).
Supervisors need policies and training for lawyers, staff and vendors (Rules 5.1 and 5.3), with tighter controls for agentic tools.
Regulator and bar guidance beyond the ethics opinions
USPTO (11 April 2024): candor, signature and confidentiality duties (37 C.F.R. parts 1, 2 and 11) apply to AI-assisted filings. Sign and review the paper yourself.
USPTO inventorship (28 November 2025): rescinds the February 2024 notice. Only natural persons can be inventors; same conception standard with or without AI.
IRS OPR Issue 2026-19: maps generative AI to Circular 230 §§ 10.22, 10.27, 10.35, 10.36 and 10.37 and IRC §§ 6713 and 7216.
SEC press release 2024-36 (2024): settled 'AI washing' charges against two advisers.
EEOC April 2024 explainer: Title VII, the ADA and related statutes apply to AI in hiring, monitoring, pay, promotion and firing. Check the current EEOC resource list.
Copyright Office AI report Part 2 (29 January 2025): copyrightability of generative-AI outputs.
Checks to carry into the matter
Confirm that the actual tool, feature and processing terms permit use of this information, including access, retention, training and disclosure conditions.
Verify the applicable jurisdiction and the version of each relevant law, rule or instrument in force at the material time.
Resolve exceptions and record the responsible lawyer’s approval of the exact output and intended use.
Limits on AI assistance
Confirm firm, client and applicable court or professional restrictions before using any AI feature.
What changes in United States
What changes in United States
Practice area
What changes here
Open
Commercial Litigation & Dispute Resolution
Open each authority in a reporter or official database Sign only what you can certify under Rule 11 Do not ask another tool whether cites are real; that does not satisfy Mata or Opinion 512
File in federal bankruptcy court Do not use Corporations Act or Bankruptcy Act 1966 (Cth) forms Open the current Code section and local rules, including any AI standing order
Apply the EEOC position (April 2024): federal discrimination laws cover AI in these activities Open the current EEOC resources and the statute Check that any 2023 EEOC document is still live
Confirm which state's family law governs Keep protected addresses, medical information and children's identifiers out of public tools Apply Rule 1.6 and California's guidance on confidentiality
Sign the response yourself, as the USPTO notice (11 April 2024) requires Review every statement: filing certifies under 37 C.F.R. § 11.18 Do not delegate the duty of candor (37 C.F.R. §§ 1.56, 11.303)
Name the statute that applies; there is no general federal privacy law Open the California AG CCPA page and the HHS HIPAA page Open Colorado SB26-189 if consequential decisions are in scope
Check the advice rests on reasonable assumptions (IRS OPR 2026-19) Authenticate citations, calculations and forecasts Keep return information out of public tools (IRC §§ 6713, 7216)
Apply PM 25-40 (8 August 2025): no ban and no mandatory AI disclosure Check the judge's standing orders Verify every citation: fake citations can lead to discipline (8 C.F.R. § 1003.102)
Keep a separate account and complete records (Rule 1.15) Confirm each disbursement yourself or through a supervised person following procedure Do not let the tool send money
Default to no for anything relating to the representation Allow input only if vendor terms, Rule 1.6 and any sector statute are satisfied Apply Opinion 512 and the Florida, California, Texas and D.C. opinions
Bill for work done (Rule 1.5 and Opinion 512) Do not bill time not spent (Florida 24-1, California) Do not mark up AI overhead without the required agreement
Check UPL, advertising and confidentiality risks (Opinion 512, Florida 24-1) Do not let an agentic tool give advice without review From 1 January 2027 (Colorado), do not present outputs as 'licensed professional services'
Apply existing FRE Articles 7 and 9 Do not cite proposed FRE 707 as in force; it was not recommended for action on 7 May 2026 Note that deepfake rules remain under study
Apply the EEOC position: discrimination laws cover AI in employment decisions Check Colorado SB26-189 duties, which start 1 January 2027 Do not certify 'compliance' from a generated checklist
Apply state law; do not import an Australian statute or form Remove any PEXA or FIRB step Apply Rule 1.6 to identity documents and contracts in any tool
Do not use an Australian SOP form Open the state lien or prompt-pay statute and the contract Verify every generated statutory reference (Formal Opinion 512)
Apply the state's tort and workers' compensation law Check the contingency fee under Rule 1.5; fees must stay reasonable when tools cut time Treat medical records in any tool as covered by HIPAA
Apply Rule 23 (class actions) and Rule 26 (discovery) Check whether protective orders forbid public tools Apply Rule 11 to filings that rely on the review
Identify the state of incorporation and open its code (Delaware or another state) Verify every section cited Review generated advice independently, as Formal Opinion 512 requires
Identify the state statute; there is no national strata Act Do not import an Australian statute or form Verify generated statutory references (Formal Opinion 512)