Patients are using AI to audit their bills, so tighten billing accuracy before claims go out, give your front desk a clear escalation path, and keep PHI out of unapproved AI tools. With CMS reporting a 10.3% improper payment rate for Medicare E/M services in 2024, accurate claims are your best defense, and Tebra’s AI Billing Assistant flags higher-risk claims before submission.

Patients have more tools than ever to question what they owe — and increasingly, those tools include artificial intelligence. Patients have started using generative AI tools like ChatGPT and Claude to review complex medical bills, decode CPT codes, and flag charges such as duplicates or possible upcoding, according to recent New York Times reporting. For private practices, this growing transparency can be a good thing, but it also raises the stakes for billing accuracy and patient communication. AI won't always get the answer right, yet it can make patients more informed — and more likely to challenge a charge they don't understand. Practices need to know what these tools can and can't reliably uncover and ensure their front-desk processes, billing workflows, and technology are ready for a new generation of AI-assisted billing questions.

Medical billing errors are mistakes on a claim or patient bill, such as duplicate charges, incorrect CPT codes, upcoding, unbundling, or charges for services not received, that can lead to denials, incorrect patient balances, or disputes

Patients are increasingly turning to ChatGPT and other AI tools to make sense of medical bills that can be confusing, expensive, and difficult to verify. AI offers a fast, accessible starting point when patients aren't sure what they owe or whether a charge is correct.

Key reasons patients are using ChatGPT to find medical billing errors include the following:

There isn’t a reliable national study showing that a specific percentage of all bills include medical billing errors. However, medical billing errors aren't uncommon. CMS reported a 10.3% improper payment rate for Medicare evaluation and management services in 2024, with incorrect coding accounting for nearly half of those improper payments. This data suggests that medical billing errors statistics are significant enough for practices to take seriously, particularly as patients gain new AI-powered tools to scrutinize charges and identify potential discrepancies.

Patients are beginning to use generative AI to scrutinize medical bills, but the trend is still emerging. AI may help patients spot discrepancies, decode billing terminology, and formulate more specific questions — even when it cannot determine on its own whether a provider's charge is correct.

For example, in one widely reported case, a family used AI tools to analyze a $195,628 hospital bill, flag questionable charges, and build a dispute that ultimately helped reduce the bill by roughly $163,000. A separate first-person account describes a patient who received an $847 bill after a routine blood test, uploaded it to ChatGPT, and says the tool helped identify a test the provider had never ordered. After the patient contacted the billing department, the charge was removed and the bill reportedly fell to $340.

As patients use AI to scrutinize their bills and identify potential medical billing errors, medical billers should expect more patients to arrive with specific questions about codes, charges, insurance adjustments, and their financial responsibility. Some AI-generated concerns will uncover legitimate medical billing errors and other issues. Other findings may be based on incomplete information or incorrect assumptions. Either way, billing teams need to be prepared to explain — and defend — the bill.

AI can help patients spot potential problems such as duplicate charges, unfamiliar services, inconsistencies between bills and EOBs, and charges that warrant a closer look. However, AI generally can't determine whether a charge is truly incorrect without access to the medical record, payer-specific rules, contracts, and other context behind the claim.

AI is well suited to spotting visible billing discrepancies and medical billing errors because it can quickly compare codes, dates, charges, totals, and information across documents such as medical bills and EOBs. It can recognize duplicates, inconsistencies, unusual patterns, and mathematical discrepancies that warrant further investigation, even if it cannot determine whether the underlying charge or code is ultimately correct.

For example, if a patient uploads a bill and an explanation of benefits (EOB), AI might notice that a patient was charged a $40 copay even though the EOB shows no copay was due for that visit. It can flag the discrepancy, but the practice would need to review the patient’s benefits and account history to determine whether the charge should be corrected. Or AI might notice that the same laboratory test appears twice on the bill, that the dates and amounts are identical, and that the EOB lists the service only once. AI can flag the second charge as a potential duplicate, but the practice would still need to verify whether both charges were appropriate.

AI is much less reliable when determining whether a medical charge is actually correct. A bill or EOB rarely contains all the information needed to make that determination, and AI typically does not have access to the patient's medical record, payer contract, benefit details, or the practice's billing system.

For example, ChatGPT might tell a patient that a level 4 office visit (99214) appears too high based on the patient's description of a brief appointment. But visit length alone doesn't determine the code: 99214 may be supported by medical decision-making or qualifying total time documented by the provider. Without that documentation, AI cannot reliably determine whether the practice coded the visit incorrectly.

AI can only analyze the information a patient provides. If a patient uploads a medical bill but not the corresponding EOB, medical record, or other relevant information, AI may lack the context needed to interpret a charge correctly — potentially flagging legitimate charges as medical billing errors or overlooking problems that aren’t apparent from the bill alone.

AI-informed patients may arrive with more detailed questions about specific codes, charges, insurance adjustments, and potential medical billing errors. This means front-desk staff must be prepared to explain basic billing information, recognize when an AI-generated concern may warrant review, and know when to escalate the question to billing or coding staff.

Instead of simply asking why they owe a balance, patients may arrive with bills they have already asked AI to analyze, along with specific codes, suspected medical billing errors, comparisons, and even AI-generated language for disputing charges. This can make billing conversations more detailed — and sometimes more challenging — as front-desk staff must distinguish between legitimate concerns and conclusions based on incomplete or inaccurate AI output.

As patients use AI to analyze their medical bills prior to paying the bill, practices may see more requests for itemized statements, CPT codes, and other details patients need to scrutinize individual charges. Making this information easy to access can reduce front-desk friction and help staff respond more efficiently when patients question what they owe.

AI can help patients quickly generate detailed questions, dispute letters, and next steps, potentially accelerating a billing concern from an initial inquiry to a formal appeal or review. Practices may have less time to resolve confusion informally, making prompt, accurate responses at the first point of contact increasingly important.

Practices can prepare for more AI-driven billing questions by strengthening billing accuracy, making charges easier for patients to understand, and giving front-desk staff clear guidance for handling and escalating disputes. 

Front-desk and billing staff should understand the types of issues and medical billing errors AI is most likely to flag so they can respond confidently when patients arrive with AI-generated questions. Training should cover how to address common concerns, verify the information, and recognize when an issue requires escalation.

Common AI-flagged issues may include:

The question for practices isn’t, ‘how can we prevent patients from using AI?’ It’s  ‘how can patients be educated to identify medical billing errors but then work with the practice to verify their concerns and resolve legitimate issues?’

Practices should establish a clear process for moving medical billing errors and questions from the front desk to the right person when additional expertise is needed. Staff should know what they can resolve themselves, what documentation to collect, and when to escalate a concern rather than debating an AI-generated conclusion with the patient.

Practices should make it easy for staff to provide itemized statements and CPT codes when patients request them, rather than requiring time-consuming manual research or multiple handoffs. Quick access to clear billing information can help staff answer questions sooner, reduce frustration, and resolve potential disputes before they escalate.

The right EHR and billing technology can help by:

Platforms like Tebra's patient payment tools include clear, branded patient statements that make charges easier for patients to review.

Practices can prevent medical billing errors by building routine checks into the billing process to catch inaccurate or inconsistent information before a bill reaches the patient. 

Verify insurance eligibility and benefits before the visit whenever possible, including coverage status and available information about copays, deductibles, and coinsurance. Accurate front-end verification can reduce downstream billing surprises and disputes over patient responsibility. Tebra's billing platform runs real-time insurance eligibility checks, so staff can confirm coverage before the visit

Make sure documentation supports the codes and services billed before the claim is submitted, and use claim edits to catch missing or inconsistent information. Addressing coding and documentation gaps upfront can prevent medical billing errors from flowing through to the patient's final bill.

Use claim scrubbing and payer-specific edits to catch missing information, coding inconsistencies, and other issues before claims are submitted. Improving clean claim rates can reduce denials, rework, and billing corrections that may ultimately create confusion for patients.

See what denials may be costing your practice. Try the free revenue recovery calculator.

Staff should avoid entering additional PHI or medical-record information into unapproved AI tools to investigate the patient’s concern, since doing so could create privacy and security risks for the practice. Protect PHI by using only vetted, HIPAA-compliant AI tools with appropriate safeguards, and establish clear rules about what patient information staff can enter into them.

HIPAA applies to covered entities and their business associates, not every company that handles health-related information. As a result, information patients share with consumer AI tools may not have the same HIPAA protections that apply when their medical practice handles their PHI.

Staff should never paste PHI — such as patient names, medical record numbers, diagnoses, treatment information, insurance details, or identifiable billing information — into public or unapproved AI tools. Practices should instead establish clear policies defining which AI tools are approved, what information staff may enter, and how patient data must be protected.

When evaluating medical billing software with AI billing tools, practices should look beyond the software’s capabilities and determine how it protects patient information, integrates with existing workflows, and supports HIPAA compliance. Is AI medical billing software HIPAA compliant? It should be. Reputable medical billing software vendors are transparent about how data is stored, used, secured, and retained.  

Look for:

Platforms like Tebra's AI Billing Assistant help billing teams identify higher-risk claims earlier, surface likely denial drivers, and focus their review to reduce denials and rework.

Also keep this in mind: When staff understand how AI contributes to minimizing medical billing errors, they can leverage it to their advantage. More specifically, the right billing technology can help practices:

Takeaway for practices: AI is giving patients a new way to scrutinize medical bills. Some of their findings will be real errors, and others will come from missing context. Either way, the practices that handle this well will have three things in place: accurate claims upfront, staff who know how to answer and escalate questions, and clear rules for protecting PHI. Start by reviewing your eligibility verification and claim scrubbing steps. Then train your front desk on the issues patients are most likely to flag.

Want to see how Tebra's AI Billing Assistant reviews claims before submission? Take the product tour.