Yes, AI can lead to false claims in two ways. When companies let AI handle billing, coding, or compliance certifications without human review, the technology can submit inaccurate claims to government programs like Medicare, and each one is a potential violation of the False Claims Act. At the same time, AI chatbots can mislead the very whistleblowers who try to report that fraud. A whistleblower attorney can help you sort a real case from an AI-generated mirage on either side of that equation.
How can AI create False Claims Act liability?
The False Claims Act doesn’t care whether a false bill came from a person or an algorithm. When healthcare providers let AI coding tools upcode Medicare and Medicaid claims without human review, a single software error could become thousands of false claims. The newest risk comes from agentic AI, or systems that act with little supervision. These tools can exceed their authorization, pull data they shouldn’t touch, or invent their own paths to a goal, and any resulting claim or certification to the government may be false.
“AI-washing” rounds out the picture: companies that overstate what their AI can do in government contracts or cybersecurity certifications may be billing for capabilities that don’t exist.
In short, if you are contracting with the government, you have to make sure your bills are accurate. You cannot delegate that job to AI, just like you cannot delegate it to a human who is not properly trained and supervised. The contracting party remains responsible for claims that are submitted on its behalf.
Are regulators paying attention to AI fraud?
Yes, even as overall enforcement has cooled. The SEC stood up its Cyber and Emerging Technologies Unit in February 2025 and put AI representations in its FY2026 examination priorities. The Department of Justice now directs prosecutors to ask how companies test, monitor, and constrain their AI systems, and its Corporate Whistleblower Awards Pilot Program pays individuals who report misconduct the government doesn’t already know about.
Whistleblower awards dipped in FY2025 (the SEC paid out $60 million, well below prior years), but AI is still a stated priority for both agencies, and the limitations periods for fraud will outlast any single administration.
What should you do if you spot AI-related fraud at work?
Document what you’ve seen in your own words, with dates and specifics. Preserve only the evidence you can lawfully access, and learn your retaliation protections before you raise concerns. Get legal advice early: the DOJ program rewards original information, so timing and sequencing (internal report or external filing, and in what order) can affect both your protection and your potential award.
Why you shouldn’t use ChatGPT to blow the whistle
Public AI tools are the wrong place to test the waters. Nothing you type into a chatbot is privileged; your queries can be stored, used for training, and in some cases turned over in litigation, potentially to the very company you’re reporting. Pasting internal documents into a public AI may also violate your confidentiality agreements with your employer and expose your identity.
The output is no safer than the input. Chatbots invent legal claims, cite statutes that don’t exist, confuse one whistleblower program with another, and inflate case values based on outlier results. And an AI-drafted complaint strips out the thing that makes whistleblower cases credible: your voice. Conversations with an attorney, by contrast, are privileged and confidential, even when they involve your employer’s information.
There is another reason not to use a chatbot to “help” you present your case: it makes our job much harder. We have to figure out what AI started with in order to get to the document you send, and it is pretty much always full of conclusions that lawyers know are not safe to make, and so somewhere between 75% and 95% of what you send may sound pretty, but it is useless to us and makes it hard to figure out what you are really saying. In a legal consultation, we want the facts from you–we will supply the law and fit the two together. Having AI do that first is like taking your raw ingredients and having AI make a cake out of it before you go the baker and order a cake. It is not helpful, it is often wrong, and it starts you off with misconceptions that we have to untangle before we can help you. Just say no to AI “legal” consultations!
Report fraud with a real advocate
If you’ve seen AI-driven fraud against the government, or you’re weighing whether what you’ve seen qualifies, Bracker & Marcus represents whistleblowers nationwide in qui tam cases, from first report through litigation. Call (770) 988-5035 or contact us online for a free, confidential consultation.