ABA Formal Op. 512.
ABA Formal Opinion 512 (2024-07-29), the ABA's first formal ethics opinion on generative AI, maps existing Model Rules onto a lawyer's use of GAI: competence (MR 1.1), confidentiality (MR 1.6), communication (MR 1.4), candor (MR 3.1/3.3), supervisory… The evidence record proves: Distinctness — outputs are measurably distinct, supporting an informed tool-selection record
Reviewed 2026-07-08
What this is.
Use. generative-AI duties · competence · confidentiality · candor · supervision.
Register. 09 · Compliance controls — one of the 17 registers of the PLANiSPHERE corpus library.
See it work.
Planisphere measures your tool deployment against ABA Formal Op. 512 and seals the result into a signed, Merkle-rooted evidence record. The grade recomputes on your own hardware; the model state never crosses the boundary.
Promise. compliance
ABA Formal Op. 512 sets the bar an AI output must clear before a tribunal or a client relies on it — reliability, candor, competence. Planisphere supplies the reliability record: a reproducible measurement of the system that produced the output, the thing the rule says you must be able to show.
Answers: How does ABA Formal Op. 512 treat AI-generated work, and what record supports review?
What it requires.
ABA Formal Opinion 512 (2024-07-29), the ABA's first formal ethics opinion on generative AI, maps existing Model Rules onto a lawyer's use of GAI: competence (MR 1.1), confidentiality (MR 1.6), communication (MR 1.4), candor (MR 3.1/3.3), supervisory responsibility (MR 5.1/5.3), and reasonable fees (MR 1.5). Load-bearing holdings: lawyers must understand the benefits and risks of the tool (which implies evaluation evidence), boilerplate engagement-letter consent is not adequate for inputting client confidences into self-learning GAI, and output must be independently verified before it is relied upon.
What the evidence record proves.
- Distinctness — outputs are measurably distinct, supporting an informed tool-selection record
- Drift — run-to-run stability is measured, supporting the duty to understand the tool's behaviour
- Coherence — internal consistency of outputs is scored
- The sealed, reproducible record is the tool-evaluation evidence behind an MR 1.1 competence file and an MR 5.3 supervision file
- Gives a firm a documented, dated basis for why it trusts a given AI tool
What this does not prove. Planisphere does not advise on confidentiality (1.6) consent, billing (1.5), or whether any specific use is ethical — those remain the lawyer's professional judgment. It measures the tool; the lawyer discharges the duty. Not legal advice and not a bar opinion.
Questions this answers.
The record this pin expects.
ABA Formal Op. 512 asks for an evidence record: ABA Formal Opinion 512 (2024-07-29), the ABA's first formal ethics opinion on generative AI, maps existing Model Rules onto a lawyer's use of GAI: competence (MR 1.1)… Read the related legal AI workflow product context at /law.
ABA Model Rules reg 09 · Rule 1.1 · 1.6 · 3.3 · 5.1–5.3 · Op 512 AI Citation Sanctions reg 09 · hallucinated-authority sanctions wave · Rule 11 / 9011 · candor-to-tribunal FRE 702 reg 09 · expert testimony · reliability gatekeeping FRE 707 (proposed) reg 09 · machine-generated evidence · Rule-702-grade reliability for AI output (PROPOSED) State Bar AI Opinions reg 09 · CA · FL · TX · NY · PA · NC bar gen-AI ethics guidance roundupSibling marks. Compliance controls
For each vertical.
ABA Formal Op. 512 for defense AI workflow
NIST AI RMF · DoW RAI · OMB M-25-21 Open the pageABA Formal Op. 512 for clinical AI workflow
HHS §1557 · HIPAA · FDA PCCP Open the pageABA Formal Op. 512 for legal AI workflow
FRE 702 · ABA Model Rules Open the pageABA Formal Op. 512 for tutor and grader workflow
Title VI · FERPA · §504 / IDEA Open the pageSee the record ABA Formal Op. 512 asks for.
The console shows the evidence; the docs show the endpoints; the briefing shows what the product does and does not claim.