0
Skip to content

Don’t Get Your Compliance Info from ChatGPT

ai apps on cell phoneAI hallucinations have appeared in more than 1,600 court filings — and counting. Here’s why your agency needs a policy that keeps generic AI out of your compliance workflow.

08/07/2026 2:00 P.M.

When a staff member types a compliance question into ChatGPT, they probably expect a helpful answer. What they may get instead is a confident, well-written, completely fabricated one. That’s the nature of AI hallucinations.

And for accounts receivable management (ARM) agencies operating under strict federal and state laws, a single misinterpreted rule or invented case citation can trigger a compliance violation, a lawsuit, or worse.

Allowing your employees to use generic AI tools to research compliance questions is “a catastrophic compliance disaster waiting to happen,” ACA CEO Scott Purcell told attendees at ACA’s 2026 Annual Convention & Expo in July.

The solution isn’t to avoid AI altogether. Rather, it’s to use the right kind of AI.

When AI Gets It Wrong in Court

The legal system is already documenting the damage. According to a June post from The National Law Journal on LinkedIn, the global database of AI hallucinations in court filings has grown from roughly 1,300 cases to more than 1,600 in just months.

In the U.S., multiple federal cases show how easily AI tools can mislead users by inventing judicial precedents and fake citations.

In both McCormick v. Texakoma Financial (U.S. District Court, Eastern District of Texas) and Barnes v. Williams (U.S. District Court, Northern District of Alabama), judges called out AI-generated legal citations and quotes that were incorrect or entirely invented. In Grimmer v. Citibank, the court went further, issuing sanctions after determining that the AI-generated citations and legal arguments did not exist in actual case law.

These cases illustrate the risks of relying on generic AI models for business-critical information, especially when the output goes unchecked. Generative AI tools predict the most statistically probable next word in a sentence, which means they can generate highly professional, authoritative-sounding citations that are entirely fictional.

For ARM professionals navigating strict federal and state laws, relying on a hallucinated rule, legal interpretation or case precedent can result in costly compliance violations or lawsuits.

The Right AI for Compliance Work

ACA built AVA specifically to address this problem. AVA is a closed-loop AI tool available to all members on ACA’s website. Unlike general AI tools, AVA does not search the open internet or draw from the broader universe of AI-generated knowledge. Its only inputs are ACA’s own human-vetted compliance resources, including ACA’s SearchPoint and Daily Decision libraries. Subscribers also get information drawn from ACA’s State Guide, with federal guides to be added soon.

Every answer AVA provides is grounded in those locked resources and includes references back to the source material.

That architecture is the point. AVA can’t hallucinate compliance information it was never given access to. The AI layer makes research faster and easier to navigate; the locked resource base ensures every answer is accurate and traceable.

As Purcell put it at the Convention: “We knew ACA members needed a shield for this, and they needed an AI tool, and that’s why we invested heavily in AVA. It is our closed-loop AI tool on ACA’s website that all of your compliance staff can use.”

Make It Policy

Knowing better tools exist is only half the equation. The other half is making sure your staff actually uses them.

A compliance team member with a quick question may just reach for the fastest tool available. If that tool is ChatGPT, the agency is exposed. The fix is a policy decision: direct compliance personnel to use vetted and trusted resources — and only those resources — when researching regulatory questions, and formalize that directive in your agency’s AI governance policy.

Every ACA member already has access to the SearchPoint and Daily Decision libraries, and one-third of our members have also subscribed to ACA’s State Guide. If your agency hasn’t incorporated these resources into your compliance workflow and your AI governance policy, the 1,600 — and growing — court filings are a clear signal that now is the time.

Remember, subscribe to ACA Daily and Member Alerts under your My ACA Assistant profile when logged in to acainternational.org.

If you have executive leadership updates or other member news to share with ACA, contact our communications department at comm@acainternational.org. View our publications page for more information and our news submission guidelines here.

One moment please...

Share Profile

This site utilizes cookies as included in our Privacy Policy and User Agreement.