ABA Formal Opinion 512: A Practical Compliance Guide

Translating ethics guidance into operational requirements for AI-using law practices.

---

On 29 July 2024, the American Bar Association Standing Committee on Ethics and Professional Responsibility released Formal Opinion 512 -- the ABA's first formal ethics guidance on generative AI in legal practice.

The 15-page opinion does not create new rules. It applies existing Model Rules of Professional Conduct to AI use, establishing that lawyers have clear obligations when deploying these tools. This guide translates the opinion's requirements into actionable compliance steps.

Scope and Authority

Formal Opinion 512 addresses generative AI tools specifically -- systems that create new content including text, analysis, and recommendations. The guidance applies to ChatGPT, Claude, legal-specific tools like Lexis+ AI and Harvey, and similar technologies.

The opinion is not binding law. Model Rules must be adopted by state bars to have regulatory force. However, as the ABA's own announcement emphasised, the opinion signals the ABA's interpretation of professional duties and will likely influence state bar guidance, malpractice standards, and disciplinary proceedings.

The Six Core Obligations

1. Competence (Model Rule 1.1)

The requirement: Lawyers must understand the benefits and risks associated with the specific AI tools they use. This understanding must be periodically updated as the technology evolves.

What this means operationally:

• Before using a new AI tool, understand how it works, what data it was trained on, and its known limitations
• Monitor for updates to tools you use regularly—capabilities and limitations change
• Recognize that "competence" in AI use is a continuing obligation, not a one-time learning event
• Uncritical reliance on AI output without appropriate verification violates this duty

Practical compliance steps:

1. Document your understanding of each AI tool you use
2. Establish a process for staying current on tool updates and published research on AI limitations
3. Create verification protocols for AI-generated content
4. Train all attorneys on AI capabilities and failure modes

2. Confidentiality (Model Rule 1.6)

The requirement: Client information entered into AI tools may be disclosed or used inappropriately. Lawyers must protect confidential information when using AI.

What this means operationally:

• "Open" AI models (those that use inputs for training) pose confidentiality risks
• Entering client data into such systems may constitute unauthorized disclosure
• Using AI tools requires understanding their data handling practices

Practical compliance steps:

1. Review the privacy policy and terms of service for any AI tool before entering client information
2. Use enterprise versions of AI tools that do not use inputs for training when working with client data
3. If you want to use client information in an open model, obtain informed consent that clearly describes the risks (including potential loss of confidentiality and intellectual property rights)
4. Establish firm policies on which tools are approved for which types of information

3. Communication (Model Rule 1.4)

The requirement: Lawyers may need to disclose AI use to clients under certain circumstances.

What this means operationally:

The opinion does not require disclosure in all cases. However, disclosure may be required when:

• The engagement agreement addresses AI use
• AI use meaningfully affects how the representation is conducted
• The client has expressed preferences about AI use
• AI use affects billing (see below)

Practical compliance steps:

1. Consider adding AI disclosure language to engagement letters
2. Discuss AI use during initial client meetings when relevant
3. Document client consent to AI use in matters where it plays a significant role
4. Be prepared to explain what AI tools were used and how if asked

4. Candor to Tribunal (Model Rules 3.1 and 3.3)

The requirement: Attorneys must carefully review AI outputs to ensure representations to tribunals are not false or misleading. Previously made false representations must be corrected.

What this means operationally:

• Every citation in a filing must be verified against primary sources
• Every factual claim generated by AI must be confirmed
• If AI-generated content turns out to be inaccurate after filing, counsel must take corrective action

Practical compliance steps:

1. Establish mandatory verification protocols for all AI-assisted filings
2. Verify every case citation exists and holds what you claim
3. Run currency checks on all legal authorities
4. Maintain documentation of verification steps
5. If errors are discovered post-filing, promptly notify the court and take corrective action

5. Supervisory Responsibilities (Model Rules 5.1 and 5.3)

The requirement: Managerial lawyers must establish firm-wide measures ensuring compliance with AI-related professional obligations. Supervisory lawyers must ensure supervised attorneys and non-lawyer assistants comply.

What this means operationally:

• Firms need AI use policies, not just individual practices
• Partners are responsible for associate and staff AI use
• Training must extend to all personnel who use AI in client matters

Practical compliance steps:

1. Develop written AI use policies covering approved tools, verification requirements, and confidentiality protections
2. Train all attorneys and staff on policies
3. Establish review procedures for AI-assisted work product
4. Document policy compliance
5. Periodically audit AI use against policies

6. Reasonable Fees (Model Rules 1.5)

The requirement: Fees charged in connection with AI use must be reasonable. Several specific principles apply.

What this means operationally:

Learning time: Lawyers may not charge clients for time spent learning a technology to be used generally—unless a client specifically requests a particular AI tool, in which case learning time for that specific tool may be billable with disclosure.

AI costs: AI tool costs may be part of overhead or charged to clients, but if charged, full explanation and informed consent are required in advance.

Time savings: If AI produces significant time savings, billing practices cannot duplicate charges or falsely inflate hours.

Practical compliance steps:

1. Establish policies on AI-related billing
2. If charging AI costs to clients, disclose and obtain consent
3. Track time accurately—bill for actual work, not what the task would have taken manually
4. Review bills for AI-assisted matters to ensure reasonableness

Compliance Checklist

Use this checklist to assess current compliance:

Competence

• Attorneys understand capabilities and limitations of AI tools used
• Verification protocols exist for AI-generated content
• Process exists for staying current on AI developments

Confidentiality

• AI tools reviewed for data handling practices before use
• Enterprise/private versions used for client information
• Informed consent obtained when required

Communication

• Engagement letters address AI use where appropriate
• Client preferences documented
• Disclosure made when AI use is material to representation

Candor

• All citations verified before filing
• All factual claims confirmed
• Currency checks performed
• Correction procedure exists for post-filing errors

Supervision

• Written AI use policies exist
• All personnel trained
• Review procedures in place
• Compliance audited periodically

Fees

• Learning time not billed inappropriately
• AI costs disclosed if charged to clients
• Time savings reflected accurately in billing

Looking Forward

The opinion explicitly notes that the duty to keep abreast of AI benefits and risks "is not a static undertaking." The ABA anticipates additional guidance as technology changes.

This means:

• Current compliance is necessary but not sufficient
• Policies and training must evolve with the technology
• State bar guidance may impose additional or different requirements
• Monitoring developments is part of the compliance obligation

Key Takeaways

• ABA Opinion 512 applies six existing Model Rules to AI use: competence, confidentiality, communication, candor, supervision, and fees
• Lawyers must understand AI tool limitations and verify all outputs—uncritical reliance on AI violates competence duties
• Client confidentiality requires understanding AI data handling; informed consent may be needed for open-model use
• Supervisory lawyers must establish firm-wide AI policies and ensure compliance
• Billing must be reasonable: no charging for general AI learning, accurate time tracking, and disclosure of AI costs passed to clients

---

For a deeper analysis of how Opinion 512 fits within the broader ethical framework for generative AI, see ABA Business Law Today's assessment of how the opinion offers a useful framework for practitioners navigating AI adoption.