ABA Formal Opinion 512: AI Competence for Lawyers

ABA Formal Opinion 512 (July 2024) is the first ABA opinion dedicated to generative AI. This article catalogs implementation patterns.

The opinion

ABA Formal Opinion 512 (July 2024):

A lawyer's duty of technology competence (Model Rule 1.1) extends to generative AI tools. Lawyers must:

1. Understand AI capabilities
2. Understand AI limitations
3. Supervise AI output

Lawyers are responsible for AI-assisted output just as they are responsible for non-lawyer assistant output.

Source: https://www.americanbar.org/content/dam/aba/publications/dispbullet/2024-july-2024/512.pdf

Implementation patterns

Competence establishment

Working patterns observed:

1. Annual AI training (minimum 4 hours)
2. Internal certifications for specific AI tools
3. Eval framework testing on known legal queries
4. Cross-lawyer mentoring

AI capability documentation

Working patterns:

1. AI tool capability matrix per vendor
2. Lawyer-facing "what AI can/can't do here" guide
3. Eval framework with statistical accuracy targets

AI limitation documentation

Working patterns:

1. Known failure modes per vendor
2. Citation accuracy benchmarks (target >95%)
3. Hallucination rate benchmarks (target <2%)
4. Periodic re-audit

Supervision requirements

Working patterns:

1. Lawyer review of every AI-assisted filing
2. Audit logs of AI use per matter
3. AI suggestion acceptance/rejection tracking
4. Quarterly audit reports

Layered compliance framework

Layer 1: Initial setup

• Vendor vetting
• BAA coverage
• Audit log infrastructure
• Compliance policy

Layer 2: Lawyer training

• Annual AI competence training
• Internal certification
• Eval framework

Layer 3: Per-matter use

• Engagement letter disclosure
• Pre-filing verification
• Audit log entry per AI query

Layer 4: Periodic review

• Quarterly eval framework
• Vendor compliance updates
• New tool evaluations

ABA Opinion 533 supplements

ABA Opinion 533 (December 2024) refines 512:

• Lawyer must advise clients about AI's role
• Can bill for AI-assisted time if disclosed
• Cannot bill purely for AI output

Source: https://www.americanbar.org/content/dam/aba/publications/dispbullet/2024-december-2024/533.pdf

Practical implementation

For a 50-attorney firm:

1. Vendor due diligence (1-2 weeks)
2. Audit log infrastructure setup (2-4 weeks)
3. Engagement letter template (1 week)
4. Training program design (1 week)
5. Pilot deployment (4-6 weeks)
6. Full deployment (1 month)
7. Quarterly audits

Total: 3-4 months.

Acknowledgments

This article summarizes public sources as of early 2026.

Dillon Deutsch has worked with ABA compliance for AI deployments. https://courtgpt.ai