AI governance for general counsel and chief compliance officers at insurers and banks
Know where your team uses AI, and whether anyone catches its mistakes.
An AI governance assessment answers both questions. You get a written report that ranks each risk and says what to fix first. When a regulator or your board asks, you will have an answer.
Led by an attorney admitted in the District of Columbia and Maryland, with 15+ years in financial services compliance.
Services
Start with the assessment
The AI governance assessment has a set scope and ends with a written report. Advisory and training can follow it or stand on their own.
First step
AI governance assessment
A look at where your team uses AI, and whether the people who review its work would catch a mistake.
What you receive
A written report that ranks each problem by risk and says what to fix first. A board member can follow it.
Illustrative first page of the report. The insurer and its figures are invented; the format is what a client receives. Every finding names a fix.Ask about an assessment
What it covers
Which AI tools your team uses, and for what
What data goes into each tool, and whether it should
How risky each use is
Who reviews each tool’s work, and whether they could catch a mistake
A scored sample of the AI’s work, using a method from my published research
Whether your outside law firms and vendors use AI on your work
How your practice compares with the rules and guidance below
The comparison uses:
The National Association of Insurance Commissioners (NAIC) Model Bulletin on the use of AI systems by insurers, as adopted in your state
American Bar Association Formal Opinion 512, on lawyers’ use of AI tools
The National Institute of Standards and Technology (NIST) AI Risk Management Framework
Federal Trade Commission guidance on AI
The European Union AI Act
The comparison is not a legal opinion.
Ongoing engagement
Advisory
Ongoing help as your AI use grows, so reviews keep working day to day.
You tell me where your team uses AI and what is at stake.
30 minutes
2
Review
I look at your AI tools and policies, then at how reviews really happen, and test a sample of the AI’s work.
On site or remote
3
Written report
The report ranks each problem by risk and says what to fix first.
Board-ready
4
Follow-through
If you want help making the fixes, the work can continue as advisory.
Optional
Why it matters
Why reviews break down when AI speeds up the work
AI can write a draft or score a risk in seconds. A person still has to review it, often with less time than before. So the review shrinks to a quick look. If the answer looks right, it goes out.
More rules rarely fix this. Many add steps without making the review any better.
Enterprise AI platforms keep a log of each question and answer. The log doesn’t show whether anyone reviewed the answer against the source.
A review works when the reviewer could actually catch a mistake. It also needs a record that shows it happened. This is what that record looks like:
Sign-off log, illustrative excerpt. Names and files are invented.
File
AI output
Reviewer
Checked against
Result
CL-2041
Record summary
R. Alvarez
Medical record, 38 pp.
Accurate
CL-2042
Record summary
R. Alvarez
Medical record, 12 pp.
Prior injury omitted. Corrected.
UW-0877
Declination letter
M. Chen
Underwriting file
Accurate
One line per review: who looked, what they compared it with, and what they found. When an examiner asks, this is the answer.
The advice draws on my published research about how AI changes professional judgment and how to test AI on legal work. The papers are posted on the Social Science Research Network (SSRN) and arXiv.
How AI models answer legal questions that leave out key facts. Written with Vincent, Yu, Bean and Seedat. Best Paper Honorable Mention and an oral presentation at the AI for Law Workshop, International Conference on Machine Learning (ICML) 2026.
I founded DANDO and do every engagement myself. I’m a regulatory attorney who came to AI from compliance work: 15+ years across anti-money laundering rules, healthcare privacy, private equity compliance and large-firm practice. Alongside this practice, I test leading AI models on legal tasks with the teams that build them.
EducationJuris Doctor, University of Connecticut School of Law. Bachelor of Arts in Criminology and Criminal Justice, University of Maryland, College Park.
Contact
Start with a 20-minute call.
No charge. You describe where your team uses AI; I say whether an assessment would help and what it would cover.
