California State Bar AI Ethics Guidance: What Attorneys Must Know

Overview

The State Bar of California's Standing Committee on Professional Responsibility and Conduct (COPRAC) has issued formal guidance on attorney use of generative AI tools in legal practice. The guidance — published as a formal ethics opinion — addresses the intersection of existing California Rules of Professional Conduct with AI-specific workflows, and does not create new rules. It interprets obligations that already bind California attorneys under Rules 1.1 (competence), 1.6 (confidentiality), 5.1 and 5.3 (supervision), and related provisions.

The opinion is significant because California has one of the largest attorney populations in the United States and because COPRAC opinions, while advisory rather than binding, carry substantial weight in disciplinary proceedings and malpractice litigation. Attorneys practicing in California — whether at large firms, solo practices, or in-house — should treat this guidance as the operative compliance baseline for AI tool use until the State Bar issues amended or superseding guidance.

Obligations Addressed

Competence (Rule 1.1)

California Rule 1.1 requires attorneys to apply the legal knowledge, skill, thoroughness, and preparation reasonably necessary for the representation. COPRAC's AI guidance interprets this to mean that an attorney who uses AI tools must have sufficient understanding of how those tools work — their capabilities, their documented failure modes, and their limitations — to supervise AI-generated output responsibly.

Practically, this means an attorney cannot simply accept AI-generated legal research, contract analysis, or drafted motions without independent verification. The guidance is explicit that hallucinated citations — fabricated case names, non-existent statutes, or misrepresented holdings — that appear in filed documents represent a competence failure, not merely a technical error. The attorney bears responsibility for the output.

Confidentiality (Rule 1.6)

Rule 1.6 prohibits attorneys from revealing client information without informed consent, with narrow exceptions. The AI guidance addresses a specific risk: when an attorney submits client documents or client-identifying information to an AI tool, that data may be retained, used for model training, or accessible to third parties depending on the vendor's data practices.

COPRAC's position is that attorneys must conduct reasonable due diligence on the data-handling practices of any AI tool before inputting client confidential information. This includes reviewing vendor terms of service, data processing agreements, and retention policies. Using a consumer-tier AI product that retains query data for training purposes — without client consent and without a data processing agreement — is treated as a potential confidentiality breach.

Supervision (Rules 5.1 and 5.3)

Rules 5.1 and 5.3 govern attorney supervision of other lawyers and non-lawyer assistants respectively. COPRAC's guidance extends the supervision framework to AI tools: when an attorney delegates a task to an AI system, they retain the same supervisory obligations they would have if delegating to a junior associate or paralegal.

This has a concrete implication for workflow design. An attorney who instructs an AI tool to draft a brief, identify relevant case law, or summarize deposition transcripts must review the output with the same critical eye they would apply to work product from a first-year associate. The guidance notes that the supervisory obligation scales with the stakes of the task — AI output in a high-stakes litigation context requires more rigorous review than AI-assisted scheduling.

Disclosure to Clients and Courts

The guidance addresses two distinct disclosure questions: disclosure to clients and disclosure to tribunals. On client disclosure, COPRAC takes the position that attorneys should inform clients when AI tools are being used in ways that could affect the representation — particularly when AI use affects billing, the handling of client data, or the nature of the work product delivered.

On court disclosure, the guidance does not impose a blanket requirement to disclose AI use in all filings. However, it notes that some California courts have issued standing orders requiring disclosure, and that truthfulness obligations under Rule 3.3 apply to representations about how work product was prepared. Attorneys should check applicable local rules and standing orders before filing AI-assisted documents.

Obligation Summary

What the Guidance Does Not Cover

COPRAC's AI opinion focuses on existing rules applied to new technology. It does not address several questions that practitioners commonly raise:

• Whether AI-assisted work product constitutes unauthorized practice of law when used by non-attorneys without attorney supervision
• How AI use interacts with attorney-client privilege, particularly when AI vendor infrastructure is involved in processing privileged communications
• Specific billing guidance beyond the general reasonableness standard in Rule 1.5
• Whether attorneys must disclose AI use in transactional work (as opposed to litigation filings)
• How the guidance applies to AI tools embedded in practice management platforms versus standalone AI research tools

These gaps are not unique to California — they reflect the pace at which bar guidance has developed relative to the actual deployment of AI in practice. Attorneys facing these questions should consult the State Bar's ethics hotline or seek a formal opinion request through COPRAC.

Relationship to ABA Formal Opinion 512

The ABA issued Formal Opinion 512 on generative AI in July 2024, addressing competence, confidentiality, communication, and supervision obligations under the Model Rules. California's guidance is broadly consistent with the ABA's framework but operates under California's own Rules of Professional Conduct, which differ from the Model Rules in several respects.

Notably, California has not adopted the Model Rules wholesale — it maintains its own rule numbering and some substantive differences. California attorneys should not assume that ABA Formal Opinion 512 maps directly onto their obligations. The COPRAC opinion is the controlling reference for California practice.

Practical Steps for California Attorneys

1. Audit current AI tool use. Identify every AI tool in use at your firm or practice and confirm whether client data is being processed through any of them.
2. Review vendor data practices. Obtain and read the data processing agreement or terms of service for each tool. Confirm whether the vendor retains query data, uses it for training, or shares it with third parties.
3. Establish a verification protocol. For any AI-generated legal research or citations, require independent verification against primary sources before the content is used in filings or advice to clients.
4. Check applicable court orders. Before filing AI-assisted documents in any California court, confirm whether the assigned judge or court has issued a standing order on AI disclosure.
5. Update client engagement letters. Consider adding disclosure language describing how AI tools may be used in the representation and how client data is protected.
6. Train supervising attorneys. Ensure that partners and supervising attorneys understand their oversight obligations when junior lawyers or staff use AI tools to generate work product.