Aspen Management Group | Bar Ethics and Compliance Aspen Management Group | Bar Ethics and Compliance | [5/26/2026]
TL;DR – Maryland, DC, Virginia AI guidance for lawyers
- Maryland: Non‑binding MSBA AI ethics guidance plus a proposed citation‑verification rule. Focus on fact‑checking AI output and protecting client data.
- District of Columbia: Formal Ethics Opinion 388 on generative AI. Maps competence, confidentiality, billing, and supervision duties directly onto AI use.
- Virginia: Legal Ethics Opinion 1901 (and proposed 1902). Addresses fees, confidentiality, supervision, and duties when AI‑generated errors appear.
Across all three jurisdictions, AI is treated as a tool under existing rules. Lawyers stay fully responsible for verification, confidentiality, reasonable fees, supervision, and candor.
What AI guidance exists today in Maryland?
Maryland attorneys are asking a fair question: “What rules do I need to follow if I use generative AI in my practice?” The answer is that Maryland does have AI‑specific guidance, but it is advisory rather than a new set of binding ethics rules.
What is the MSBA’s AI ethics guidance?
In 2025, the Maryland State Bar Association’s Artificial Intelligence Task Force released an advisory document titled “An Overview of Ethical Considerations for Attorney Use of Generative Artificial Intelligence Technologies.”
This guidance:
- Stresses that generative AI output must be independently reviewed and fact‑checked by the attorney before it is relied on.
- Emphasizes existing duties of competence, confidentiality, reasonable fees, and candor in the context of AI tools.
- Warns about risks such as hallucinated facts, incorrect citations, and inadvertent disclosure of sensitive client information when using third‑party AI platforms.
It is important to note that this MSBA guidance is non‑binding. It does not replace the Maryland Lawyers’ Rules of Professional Conduct. Instead, it explains how those existing rules apply when attorneys choose to use generative AI.
What’s the risk with citation verification and fake authorities?
Separately, in March 2026, a top Maryland rules committee voted to require lawyers to verify every case citation and explicitly target fake AI‑generated authorities.
Under the proposed rule:
- Attorneys would need to certify that they have confirmed the accuracy and existence of each citation they submit.
- Willful use of fake or non‑existent citations could lead to monetary sanctions, reinforcing existing duties of candor to the tribunal.
At the time of writing, this citation‑verification requirement reflects a committee recommendation that still needs formal adoption by the state’s highest court. It builds on existing obligations rather than creating an entirely new “AI rule.”
What does this mean in practice for Maryland attorneys?
Putting this together, Maryland’s message to attorneys is clear:
- Generative AI can be used, but it must sit inside the normal ethics framework.
- Lawyers remain fully responsible for the accuracy, relevance, and appropriateness of anything they submit or rely on, whether it came from a human draft or an AI tool.
- Confidentiality and data‑protection duties apply to any client information used with AI, particularly public or third‑party tools.
- Courts and rulemakers are increasingly explicit that fake citations and unchecked AI output are not acceptable and may be sanctioned.
For Maryland attorneys, the practical takeaway is simple: generative AI may assist with research, drafting, and summarization, but it never replaces the lawyer’s obligation to verify, to exercise judgment, and to protect clients.
What does DC’s Ethics Opinion 388 say?
In April 2024, the DC Bar issued Ethics Opinion 388, “Attorneys’ Use of Generative Artificial Intelligence in Client Matters.”
Key points:
- Competence (Rule 1.1)
Lawyers must have a reasonable and current understanding of how a generative AI tool works before relying on it. Technological ignorance is not an excuse. - Confidentiality (Rule 1.6)
Attorneys must be careful about what client information is entered into AI tools and must understand how vendors store, train on, and use that data. - Candor and accuracy
AI output must be verified. Filing hallucinated cases or incorrect analysis exposes the lawyer to the same sanctions risk seen in other jurisdictions. - Disclosure
DC does not impose a blanket duty to disclose AI use. Opinion 388 ties disclosure to tribunal rules and the duty of candor rather than a categorical mandate.
In practice, DC treats AI as a supervised assistant. The lawyer is responsible for understanding the tool, protecting client data, and checking all output before use.
What do Virginia’s Legal Ethics Opinions 1901 and 1902 say?
Virginia has moved forward with formal AI ethics guidance.
What does Legal Ethics Opinion 1901 say about generative AI and reasonable fees?
Virginia’s LEO 1901 directly addresses generative AI use by Virginia attorneys.
Highlights:
- Competence (Rule 1.1)
Lawyers must understand AI tools well enough to use them effectively and to spot errors and hallucinations. - Supervision (Rule 5.3)
AI is treated like a non‑lawyer assistant: it must be supervised, reviewed, and cannot be allowed to “practice law” on its own. - Fees (Rule 1.5)
The opinion clarifies that using AI does not automatically require cutting fees to actual time spent. Reasonable non‑hourly fees can reflect efficiency gains, skill in using technology, and the value of the services.
At the same time, AI tool costs and pass‑through expenses require client consent and transparent billing. - Confidentiality (Rule 1.6)
Lawyers must make reasonable efforts to safeguard client information when using AI, including vendor due diligence (training practices, retention, sharing, and contractual protections).
What does proposed Legal Ethics Opinion 1902 say about AI-generated errors?
Virginia is also considering LEO 1902, which addresses duties when opposing counsel files AI‑generated errors.
It emphasizes that failing to verify AI outputs may violate several rules, including:
- competence,
- diligence,
- meritorious claims,
- candor, and
- honesty
LEO 1902 guides lawyers to use professional judgment in deciding whether to notify the court or disciplinary authorities, and discourages using bar complaints purely for tactical advantage.
LEO 1901 has been approved by the Virginia State Bar Council and sent to the Supreme Court of Virginia for approval. LEO 1902 is out for public comment.
What’s the bottom line for DC, Maryland, and Virginia attorneys?
Across Maryland, DC, and Virginia, the message is consistent:
- AI can assist, but lawyers remain fully responsible for verification, confidentiality, fees, supervision, and candor.
- Generative AI is treated as a powerful but supervised tool under existing professional conduct rules, not as a substitute for legal judgment or ethical obligations.
Aspen Management Group works with boutique advisory firms to clarify key workflows, layer in AI where it truly adds value, and build governance and training around that change.
aspenmg.net
Scott spent 20 years running a managed IT services practice serving professional services firms across the DC Metro area, and has worked in technology for 35 years. AMG helps boutique professional services firms get practical value out of AI.