An Arizona personal injury firm using AI-generated demand letters had two causation problems hiding in plain sight: unrelated medical visits counted as part of the claim, and prior injuries the AI never flagged. Here's how independent AI output audit caught both before the demands went out.
Case at a Glance
The Situation: AI Had Connected the Wrong Dots
A personal injury firm in Arizona had adopted an AI drafting tool to generate demand letters and medical chronologies across its caseload, aiming to move cases through intake and demand preparation faster. The AI tool performed well on the basics — organizing treatment records, drafting narrative summaries, and calculating damages totals from the medical billing provided.
But when the firm began routing its AI-generated demand letters through FactAssess' AI Output Audit service, a pattern emerged across multiple case files: the AI tool was making causation errors that weren't visible on a surface read, but that would have been immediately exploitable by defense counsel or an insurance adjuster.
What the Audit Uncovered
Unrelated Medical Visits Counted as Related Treatment
In several files, the AI-generated chronology and demand letter included treatment visits that had no connection to the incident at issue — unrelated medical appointments that happened to fall within the same general timeframe as the injury but addressed separate, pre-existing, or unconnected health issues. Because the AI tool was working primarily from date proximity and record volume rather than a true clinical causation analysis, these unrelated visits were folded into the treatment timeline and, in some cases, into the damages calculation itself.
Prior Injuries Not Flagged
In multiple case files, the underlying medical records referenced prior injuries to the same body region as the current claim — the exact kind of detail defense counsel actively searches for to argue that a client's condition pre-existed the incident, not resulted from it. The AI-generated drafts didn't surface or flag these prior injuries at all, leaving the demand letter's causation argument built on an incomplete picture of the client's medical history.
Why These Errors Are So Costly
Both error types share a common thread: they're invisible on a quick review but devastating in negotiation or litigation. A demand letter that unknowingly bundles unrelated treatment into a damages total is vulnerable to a defense argument that overstates the claim — and once an adjuster catches one inflated or unsupported figure, it undermines confidence in every other number in the letter. A demand letter that fails to proactively address a documented prior injury hands the defense a ready-made causation dispute instead of getting ahead of it.
Neither error is a fabrication in the way a hallucinated citation is — the underlying records are real. The failure is one of clinical and legal judgment: correctly distinguishing related from unrelated treatment, and recognizing when a prior injury needs to be addressed head-on rather than left for the defense to discover.
How the Audit Process Caught It
FactAssess' AI output audit process verifies every factual claim in a demand letter or medical chronology against the underlying source records — not just for accuracy, but for relevance and completeness. In this engagement, our reviewers:
✓
Cross-referenced every treatment date and provider visit against the documented mechanism of injury, identifying visits unrelated to the claimed incident
✓
Reviewed the full medical history in each file specifically for prior injuries, prior treatment, or pre-existing conditions affecting the same body region as the current claim
✓
Flagged both categories of issue with specific, correctable guidance — removing unrelated visits from the damages calculation and drafting proactive language addressing the prior injuries directly, rather than leaving them for the defense to raise first
The Outcome
The firm received audited, corrected demand letters across the affected files — with unrelated treatment removed from damages calculations and prior injury disclosures addressed proactively as part of the causation narrative, rather than omitted. Each corrected document came with an audit trail memo documenting exactly what was reviewed, flagged, and revised.
For a firm relying on AI-generated demand letters across a growing caseload, this engagement surfaced a systemic pattern — not an isolated mistake — giving the firm confidence that similar causation and prior-injury issues would be caught on future files as well.
Why This Matters for Every AI-Generated Demand Letter
Unrelated visits and unflagged prior injuries are two of the most common — and most consequential — causation issues we see in AI-generated demand letters, precisely because they don't look like errors on the surface. The medical records are real, the dates are real, and the AI tool isn't fabricating anything. It's simply not applying the clinical and legal judgment needed to distinguish related from unrelated treatment, or to recognize when a prior injury changes the causation analysis.
This is exactly the gap independent, source-traceable human verification is built to close — catching not just what's wrong, but what's missing.
Case Study Disclaimer
This case study is based on the client scenario supplied for this webpage. Client, claimant, attorney, and matter-identifying details have been omitted or generalized for confidentiality. The findings described are presented as the reported results of the audit and are not intended to imply that every AI-generated demand will contain the same issues. This case study does not constitute legal advice or a guarantee of audit results.
Don't let unrelated visits or unflagged prior injuries weaken your next demand. Get it audited first.
Our Services
About Us
Get in Touch with Us!
© 2026 FactAssess. All rights reserved. Privacy Policy