Tara Ravi Quoted in Part B News on Key AI Concerns in Healthcare

Bradley partner Tara Ravi was quoted in Part B News emphasizing that AI can create legal and compliance exposure even when it appears to improve efficiency or documentation quality.

“The government side is set up so that if you send them all ‘perfect’ claims, they may start auditing based on pattern recognition,” Ravi warned, noting that overly polished AI-assisted billing submissions can backfire.

Government payers may become suspicious when documentation appears unnaturally standardized or uniformly optimized. AI can unintentionally create records that look cloned or mechanically generated.

Ravi said regulators worry about “documentation that appears excessively standardized, repetitive, cloned or disconnected from the individualized patient encounter.” In practice, this means providers must ensure records still reflect unique patient interactions rather than formulaic AI-produced language.

Ravi cautioned that AI editing tools can cross legal lines if they alter the substance of a clinical encounter. “AI can help identify missing docu­mentation elements or make language clearer, but if AI-generated edits materially alter what occurred during the encounter, overstate complexity, add findings that were not documented or make the record appear more supportive of billing than the facts allow, that can create medical-record integrity and False Claims Act risk.”

She also recommended intensive human oversight when organizations first deploy AI scribes: “Make sure that you double-scribe — have a person come in and old-school scribe, then cross-compare across visits; maybe quarterly, maybe daily in the implementation phase.”

“Review may become more risk-based and audit-oriented rather than encounter-by-encounter. But organizations will still need governance structures, validation protocols and human accountability,” she added.

Ravi also stressed that providers should continuously monitor AI-assisted workflows because regulators already scrutinize repetitive documentation patterns. “Enforcement actions have historically focused on allegations involv­ing cloned, repetitive or insufficiently individualized documentation, and AI-assisted workflows could increase that risk if not properly monitored. A possible way to get around that is to show that you’ve contin­ued to oversee the product: Here are our compliance parameters, we’ve used a human to review 60 of our claims every quarter, etc.”

The full article, “AI Chatbots Create Medico-Legal Trouble; Reduce Your Risk of Harm,” was published by Part B News on May 18, 2026. (login required)