Skip to main content
  1. AI Legal Ethics by State/
  2. State AI Ethics Rules for Attorneys/

California AI Ethics Rules for Attorneys

Table of Contents

California was the first state to approve regulatory guidance for attorney use of generative AI, releasing its “Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law” in November 2023. The California State Bar has characterized this guidance as “guiding principles rather than best practices,” reflecting the rapidly evolving nature of AI technology.


Key Guidance Document
#

Practical Guidance for Use of Generative AI (November 2023)
#

The California State Bar’s Standing Committee on Professional Responsibility and Conduct (COPRAC) released this comprehensive guidance addressing AI use across the full spectrum of legal practice.

Document Status: Living document intended to evolve with technological advancements

Full Document: California State Bar Generative AI Guidance (PDF)


Core Ethical Obligations
#

Confidentiality
#

Critical Requirement
Attorneys must ensure AI systems protect client confidential information before inputting any client data.

California’s guidance requires attorneys to:

  • Consult with IT or cybersecurity professionals to verify AI systems adhere to “stringent security, confidentiality, and data retention protocols”
  • Review Terms of Use to understand how inputted information is processed
  • Verify the platform does not share inputted information with third parties
  • Confirm data is not used to train or improve the AI product without consent

Under California Rules of Professional Conduct Rule 1.6, disclosure of confidential information to AI systems without adequate protections may violate the duty of confidentiality.

Competence and Diligence
#

California’s guidance establishes a high bar for competent AI use:

Verification Requirements:

  • Lawyers must “critically review, validate, and correct both the input and the output” of generative AI
  • Verification extends beyond detecting false results, it requires ensuring content “accurately reflects and supports the interests and priorities of the client”

Professional Judgment:

  • “A lawyer’s professional judgment cannot be delegated to generative AI”
  • Judgment “remains the lawyer’s responsibility at all times”

Over-Reliance Warning:

  • Lawyers should avoid over-reliance on AI “to such a degree that it hinders critical attorney analysis fostered by traditional research and writing”

Billing for AI-Assisted Work
#

California’s guidance addresses the billing implications of AI efficiency:

Permitted Billing:

  • Time spent refining AI inputs or prompts
  • Time spent reviewing and editing AI outputs
  • Actual time creating work product using AI assistance

Prohibited Billing:

  • “A lawyer must not charge the client the time saved due to the use of GenAI”
  • Billing for hours not actually worked violates fee reasonableness requirements

Duty to Supervise
#

Law firm partners and managers have affirmative supervision duties:

  • Establish clear policies regarding permissible AI uses
  • Provide training on ethical and practical aspects of AI use
  • Make reasonable efforts to ensure firm lawyers and non-lawyers comply with professional obligations
  • Create measures that give “reasonable assurance” of compliance

California Rules of Professional Conduct Implicated
#

RuleObligationAI Application
Rule 1.1CompetenceUnderstand AI capabilities/limitations; verify all outputs
Rule 1.3DiligenceDon’t allow AI use to delay or harm client matters
Rule 1.4CommunicationDisclose AI use when material to representation
Rule 1.6ConfidentialityProtect client data in AI systems
Rule 1.5FeesDon’t bill for time saved by AI; ensure reasonable fees
Rule 3.3CandorVerify all AI-generated content before court submission
Rule 5.1Supervisory DutiesEstablish AI policies; train lawyers and staff
Rule 5.3Nonlawyer AssistanceSupervise AI as you would a nonlawyer assistant

California Court AI Orders
#

Noland v. Land of the Free (Cal. Ct. App. September 2025)
#

California courts have begun sanctioning attorneys for AI hallucinations. In this landmark case:

Facts:

  • Attorney used ChatGPT, Claude, Gemini, and Grok to draft appellate briefs
  • 21 of 23 case quotations in plaintiff’s opening brief were fabricated
  • Attorney submitted multiple briefs with fake citations even after warnings

Consequences:

  • $10,000 sanctions
  • Reported to the California State Bar for potential discipline
  • Required to show opinion to client and certify compliance
  • Lost the underlying appeal

Significance: First state Court of Appeal opinion sanctioning a lawyer specifically for AI-generated hallucinations.


AI Disclosure Requirements
#

California has not mandated AI disclosure in all circumstances. Disclosure depends on materiality:

Disclosure May Be Required When:

  • AI use materially affects the representation
  • Client specifically inquires about AI use
  • Confidential information will be inputted into AI systems
  • AI use affects fee arrangements

Disclosure Not Automatically Required:

  • Routine use of AI for administrative tasks
  • AI-assisted research that is independently verified
  • Document formatting or proofreading assistance

Practical Compliance Steps for California Attorneys
#

California AI Compliance Checklist

Before Using AI:

  1. Review AI platform Terms of Use and privacy policies
  2. Consult IT/cybersecurity on data protection adequacy
  3. Verify client data won’t be shared or used for training
  4. Establish firm-wide AI use policies

During AI Use: 5. Never input client confidential information without protections 6. Maintain human judgment in all substantive decisions 7. Document your verification process

After AI Generates Content: 8. Independently verify all citations in Westlaw/Lexis 9. Check quoted language against original sources 10. Shepardize/KeyCite all cited authority 11. Review for logical consistency and accuracy

For Billing: 12. Bill only for time actually spent 13. Don’t bill for hours saved by AI efficiency 14. Disclose AI-related costs to clients

For Supervision: 15. Train all lawyers and staff on AI policies 16. Require verification before any AI content is filed 17. Establish quality control checkpoints


Malpractice Insurance Considerations
#

California attorneys should review their malpractice policies for AI-related coverage:

  • Some policies may exclude AI-related claims
  • Others may require disclosure of AI use
  • Premium adjustments may apply for heavy AI users
  • Document your verification procedures for defense purposes

Frequently Asked Questions
#

Does California require disclosure of AI use to clients?

Not automatically. California’s guidance treats AI disclosure as dependent on materiality. Disclosure is required when AI use materially affects the representation, when the client inquires, when confidential information will be inputted, or when AI use affects fees. Routine, verified AI use for research or drafting doesn’t require automatic disclosure.

Can I use ChatGPT for legal research in California?

Yes, but with verification requirements. California permits AI use for legal research, but attorneys must independently verify all citations, quotes, and legal propositions before relying on them. The Noland case demonstrates that failing to verify, resulting in fabricated citations, can lead to sanctions and bar referral.

How should California attorneys bill for AI-assisted work?

Bill only for time actually spent. You may charge for time refining prompts, reviewing outputs, and editing AI-generated content. However, you cannot bill for time saved by AI efficiency, if AI completes a task in 15 minutes that would have taken 2 hours manually, you bill 15 minutes, not 2 hours. Discuss AI fee arrangements with clients upfront.

What are the confidentiality requirements for using AI in California?

Before inputting any client information, verify the AI platform has adequate security protections, doesn’t share data with third parties, and doesn’t use inputs for training. Review Terms of Use carefully. Consider consulting IT or cybersecurity professionals. When in doubt, don’t input confidential information.

Resources
#


Questions About AI Ethics Compliance in California?

California was the first state to address attorney AI use, and its living guidance continues to evolve. Understanding your obligations under California Rules of Professional Conduct is essential for compliant AI integration in your practice.

Consult a Legal Ethics Attorney

Related

Florida AI Ethics Rules for Attorneys

On January 19, 2024, the Florida Bar Board of Governors unanimously approved Ethics Opinion 24-1, providing guidance on the ethical use of generative artificial intelligence in legal practice. Florida was among the first states to issue formal AI ethics guidance, and Opinion 24-1 has been recognized as a model for other jurisdictions.

New York AI Ethics Rules for Attorneys

New York has developed one of the most comprehensive frameworks for AI ethics in legal practice. In April 2024, the New York State Bar Association (NYSBA) released its Task Force on Artificial Intelligence report, and the NYC Bar Association issued Formal Opinion 2024-5. Together, these documents provide extensive guidance for New York attorneys using AI.

Pennsylvania AI Ethics Rules for Attorneys

In May 2024, the Pennsylvania Bar Association and Philadelphia Bar Association jointly released Formal Opinion 2024-200, providing comprehensive guidance on ethical issues regarding attorney use of artificial intelligence. This joint opinion reflects collaboration between the state’s two major bar associations and addresses the full range of AI ethics considerations.

Texas AI Ethics Rules for Attorneys

In February 2025, the Professional Ethics Committee for the State Bar of Texas issued Opinion 705, providing comprehensive guidance on Texas attorneys’ use of generative artificial intelligence. This opinion builds on the work of the Taskforce for Responsible AI in the Law (TRAIL), an initiative launched by the Texas State Bar’s Immediate Past President, Cindy Tisdale.