AI-Drafted Demand Letters: Ethics for Michigan PI Lawyers
The demand letter is the document that often sets the value of a personal injury case. It is also one of the first tasks lawyers now hand to generative AI. A tool can read a medical summary and produce a polished demand in minutes, which is a real efficiency. It can also invent a treatment that never happened, miscite a statute, or quietly send a client’s confidential records to a server the lawyer never vetted. The technology is allowed in Michigan, but the responsibility for what goes out the door stays entirely with the lawyer. This guide explains how AI is used to draft demands and the ethical duties that govern that use.
How AI Is Used to Draft Demands
Personal injury firms increasingly feed an AI tool the medical records, bills, wage-loss documentation, and liability facts, then ask it to assemble a demand letter: a narrative of the incident, a summary of injuries and treatment, an itemization of damages, and a settlement demand. Drafting correspondence and documents is now among the most common uses of generative AI in law. Used well, the tool drafts faster than a human and frees the lawyer for analysis and negotiation.
The risk is that a demand letter is an advocacy document built on facts. If the underlying facts are wrong, the letter is worse than useless. A generative model predicts plausible text; it does not know whether a date, a diagnosis, or a dollar figure is true. A demand that overstates treatment or misstates the record hands the adjuster, and later the defense, a credibility weapon that can sink the case.
The Governing Ethics Rule: ABA Formal Opinion 512
On July 29, 2024, the ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512, its first formal guidance on generative AI. The opinion does not create new rules. It applies the existing rules of professional conduct to AI, and it identifies the duties that matter most: competence, confidentiality, communication with the client, candor, supervision, and reasonable fees. Michigan lawyers operate under the Michigan Rules of Professional Conduct, which mirror the model rules the opinion interprets, so the analysis maps directly onto Michigan practice.
Five Duties That Control AI Demand Drafting
1. Competence and verification (MRPC 1.1)
Competence now includes a working understanding of the tools a lawyer uses and their limits. A lawyer who uses AI to draft a demand must understand that the tool can fabricate, and must independently verify every material fact and citation before the letter goes out. Every diagnosis, date of treatment, bill amount, and legal reference must be checked against the actual record. The lesson from sanctioned lawyers who filed AI-invented case citations applies with equal force to a demand letter, as our guide to AI hallucinations and court sanctions explains.
2. Confidentiality (MRPC 1.6)
A demand is built from the client’s most sensitive material: medical records, mental-health history, and financial data. Feeding that information into a public AI tool can disclose it to the vendor and, depending on the terms, expose it to training or storage the lawyer cannot control. Before any client information goes into a tool, the lawyer must understand how the vendor handles, stores, and uses the data, and must use a tool whose terms protect confidentiality. The privilege and disclosure concerns parallel those in our discussion of AI note-takers and privilege risks.
3. Communication with the client (MRPC 1.4)
Depending on the circumstances, a lawyer may need to tell the client that AI is being used in the matter and obtain informed consent, particularly where confidential information will be put into the tool. The client’s reasonable expectations about how their information is handled are part of the communication duty.
4. Candor and accuracy (MRPC 3.3)
While a demand letter goes to an adjuster rather than a court, the duty not to make false statements of material fact runs through all of a lawyer’s work. A demand that knowingly or recklessly overstates injuries or fabricates treatment is both an ethical problem and a strategic disaster. AI raises this risk because fabricated content can look entirely credible.
5. Supervision (MRPC 5.3)
An AI tool functions like nonlawyer assistance, and the lawyer must supervise its output as such. That means putting office policies, training, and review steps in place so that nothing a tool produces reaches a client, an adjuster, or a court without a competent lawyer’s independent check.
| DUTY | MICHIGAN RULE | WHAT IT REQUIRES WITH AI DEMANDS |
|---|---|---|
| Competence | MRPC 1.1 | Understand the tool’s limits; verify every fact and citation |
| Confidentiality | MRPC 1.6 | Vet the vendor; protect client records before any input |
| Communication | MRPC 1.4 | Inform the client and obtain consent where appropriate |
| Candor | MRPC 3.3 | No false or fabricated statements of material fact |
| Supervision | MRPC 5.3 | Treat AI output like nonlawyer work; review before it goes out |
A Safe Workflow for AI-Assisted Demands
AI can draft demand letters in Michigan without crossing ethical lines, but only inside a disciplined process:
- Use a vetted, confidential tool. Confirm in writing how the vendor stores and uses input data before any client record is entered, consistent with MRPC 1.6.
- Verify every fact against the record. Check each diagnosis, treatment date, and dollar figure in the draft against the underlying medical and billing records. Treat the AI draft as an unverified first pass.
- Check every legal reference. Confirm that any statute or case cited in the demand actually exists and says what the letter claims.
- Keep the lawyer’s judgment in the loop. Strategy, valuation, and tone are legal judgments the lawyer must own; the tool drafts, the lawyer decides.
- Document the review. Keep a record that a lawyer verified the demand before it was sent, which supports both the supervision duty and the file.
ਅਕਸਰ ਪੁੱਛੇ ਜਾਂਦੇ ਸਵਾਲ
Can Michigan lawyers use AI to draft demand letters?
Yes. There is no rule prohibiting it, and drafting correspondence is one of the most common legal uses of generative AI. The lawyer remains fully responsible for the content under the Michigan Rules of Professional Conduct and the framework of ABA Formal Opinion 512.
What is the biggest risk with an AI-drafted demand?
Inaccuracy. A generative model can fabricate a diagnosis, a treatment date, a bill amount, or a legal citation that looks real. An unverified demand can hand the defense a credibility weapon, so every material fact must be checked against the record before the letter is sent.
Is it a confidentiality problem to put medical records into an AI tool?
It can be. Under MRPC 1.6, a lawyer must protect client information. Entering records into a tool that stores or trains on input data, or whose security the lawyer has not vetted, can breach confidentiality. Use only a tool whose terms protect the data, and understand how it handles input first.
Do I have to tell my client I used AI?
Often yes. Depending on the circumstances, the communication duty under MRPC 1.4 can require informing the client and obtaining consent, especially when confidential information will be entered into the tool. ABA Formal Opinion 512 discusses when disclosure is appropriate.
What does ABA Formal Opinion 512 say?
Issued July 29, 2024, it is the ABA’s first formal ethics guidance on generative AI. It applies existing rules, competence, confidentiality, communication, candor, supervision, and reasonable fees, to lawyers’ use of AI tools rather than creating new ones.
Does using AI let me charge less or more?
Fees must remain reasonable. A lawyer generally cannot bill a client for hours the AI saved as if the work were done by hand, and cannot pass through tool costs unfairly. Reasonable-fee principles continue to apply, as Opinion 512 notes.
Building a demand letter that holds up?
Free, confidential review with Attorney Manny Chahal. We verify every fact and citation before a demand goes to the adjuster. Consultations available in English, Hindi (हिंदी), and Punjabi (ਪੰਜਾਬੀ).
ਮੁਫ਼ਤ ਕੇਸ ਮੁਲਾਂਕਣ ਲਈ 1-844-MCHAHAL 'ਤੇ ਕਾਲ ਕਰੋ