Ariel C. Brownstein

For years, healthcare payers have treated note cloning—the practice of copying and pasting electronic health record (EHR) text—as a primary red flag in fraud, waste, and abuse (FWA) investigations. Today, as the industry races to embrace Artificial Intelligence (AI) for documentation, the threat of "cloning" is not disappearing; it is simply evolving. For insurance carriers facing healthcare fraud costs estimated to exceed $400 billion annually in the U.S., understanding this new and evolving technological risk is paramount to effective claims denial and successful defense litigation

The core issue with cloned documentation is its immediate challenge to the medical necessity of billed services. When medical records contain identical or near-identical entries across multiple dates of service, the documentation cannot support the premise that unique, individualized care was provided at each encounter. This practice undermines the credibility of the entire record.

Traditional copy-and-paste charting, where clinicians simply copy-forward prior entries or borrow from templates, was quickly identified by the Centers for Medicare & Medicaid Services (CMS) and the Office of Inspector General (OIG) as a priority for audit and enforcement. Its misuse often results in a form of fraud known as up-coding—the insertion of false or irrelevant details to justify a higher, more expensive level of service than was actually rendered. Simply put, manufactured records support inflated billing.

Cloning 2.0: AI and the New Red Flags

The rapid adoption of AI-assisted documentation tools presents carriers with a new, but strikingly familiar, compliance pitfall. Just as a keyboard shortcut once generated a suspiciously repetitive note, a sophisticated machine learning algorithm can now produce a grammatically flawless but equally generic summary.

Insurance carriers must equip claims auditors with a new playbook for identifying these high-tech red flags:

• Repetitive and Boilerplate Phrasing: Like cut-and-paste, AI tools tend to reuse stock language verbatim—for instance, identical descriptions of a patient's presentation across many different encounters. The presence of uniform, verbose, or overly formal language that clashes with an experienced auditor's knowledge of a physician's typical "voice" should raise suspicion. These generic statements does not reflect individual patient encounters, creates the assumption that the narrative was manufactured to support, higher E/M coding and supports the appearance of a systematic inflation by a provider, not an isolated error.
• Overly Complete Documentation: A hallmark red flag for potential upcoding is extreme documentation thoroughness. Unlike human clinicians, who focus on relevant positives and negatives, AI systems frequently generate exhaustive, boilerplate reviews of systems. Such documentation can misrepresent the scope of the encounter, creating the appearance of higher-level services and automatically inflating the reported E/M code—despite no corresponding increase in clinical work. An example of this would be a patient presenting with a sore throat and congestion, but the note documents a 14-system Review of Symptoms (ROS), all marked negative. A routine upper respiratory complaint does not clinically justify a full multi-system ROS. This level of detail artificially supports a higher E/M level without corresponding medical necessity.
• Internal Inconsistencies: Because AI relies on patterns, it can fail to reconcile contradictory information or carry forward fabricated or outdated details. For instance, one section of an AI-generated note might state "no extremity pain," while another later mentions "episodes of upper extremity discomfort". These internal contradictions are destructive to a record's credibility and are prime targets for counsel in deposition.
• Metadata Trails: Crucially, the technology that enables AI documentation also leaves an audit trail. Carriers must leverage the power of discovery to review system logs and timestamps that reveal when AI tools were used to generate text. This metadata can prove the extent of a provider's reliance on automated shortcuts, flagging instances of potential overreliance.

Fighting Fire with Fire: The Carrier's AI Defense

The growing sophistication of provider fraud demands that insurance carriers evolve beyond static, rules-based fraud detection to advanced analytical models. The best defense against AI-driven fraud is often the strategic use of defensive AI.

• Carriers must transition to modern FWA prevention strategies by:
• Pre-Payment FWA Preventive Analytics: Moving beyond traditional post-pay audits, carriers are now leveraging machine learning models to score and flag claims for high-risk behavior before adjudication. This shift prevents the improper payment from ever being made.
• Leveraging Natural Language Processing (NLP): NLP is essential for analyzing the unstructured data in medical records, specifically clinical notes. These tools can scan millions of provider notes to detect the subtle anomalies that human auditors might miss, such as:
  • Identification of repetitive and cloned phrases across a provider's patient roster.
  • Flagging medical codes that do not align with the narrative diagnosis or description in the note.
• Predictive Behavioral Modeling: AI systems can track a provider's historical billing and documentation patterns, automatically identifying statistically significant deviations from their peers. When a provider suddenly increases their volume of complex E/M codes (a classic up-coding indicator) or exhibits unusual service combinations, the system flags the provider as a high-risk outlier for focused investigation.
• Network Link Analysis: Advanced analytics can uncover collusive networks of providers who might be sharing patients or services to perpetrate fraud.

In conclusion, the ultimate lesson for carriers is that documentation is not merely about filling space; it is about telling the patient's distinctive and current story. Anything—whether a copy-paste command or a machine learning algorithm—that dilutes that unique story and creates repetitive or over-documented records is a pathway to claims failure and potential fraud. Insurance carriers must treat AI documentation with the same rigorous scrutiny once reserved for chart cloning, updating audit protocols to focus on individualized clinician attestation, customization, and metadata that reveals overreliance on automation.