#14B Waiting for Godot in Healthcare AI: Part II

Read Waiting for Godot in Healthcare AI: Part I.

As Day 2 dawns, Vladimir enters to find the sparse tree now growing new leaves. Estragon joins him, arriving barefoot, only to reveal that he was beaten overnight.

The once powerful master, Pozzo (now blind) and his long-abused servant Lucky (now mute) return.

Estragon tries (in vain) to remember the events of the prior day. Vladimir sings a folk song about a dog being beaten to death for stealing a piece of bread.

Their grim wait for Godot grinds on…

The struggle for AI business sector supremacy leaves even the powerful EHR duopoly and their big U.S. health system customers waiting — and their patients simply making do.

Concentrating markets hear the fervent calls for “more government regulation,” another “oversight committee," a scientific “moratorium”, but they do not respond.

So regulatory oversight must be applied to protect the consumer from unfairly high prices and/or unsafe goods and services. 

Regulatory Oversight

Antitrust laws and regulations are designed to prevent the concentration of economic power (often occurring through M&A) as oligopolies (or monopolies or duopolies) that can foster other anticompetitive practices (price fixing) that negatively impact consumers.  

The U.S. Department of Justice (DOJ) Antitrust Division is responsible for applying antitrust laws so that free markets do not become oligopolies because of corporate mergers and acquisitions [M&A] (see DOJ Merger Guidelines § 2.1, 2023). M&A transactions that increase the HHI by >100 points in highly concentrated markets or that result in >30% market share for the combined firm are viewed as being likely to enhance market power and are subject to greater DOJ antitrust regulatory scrutiny.

The Hart-Scott-Rodino Act (HSR Act, 1976; updated 2024) mandates reviews of consolidations >$133.9M per person or >$535.5M per deal. Under the HSR Act, both the DOJ and Federal Trade Commission (FTC) reviewed the $28.3B acquisition of the Kansas City-based Cerner by Oracle regarding anti-competition concerns. Prior to this June 2022 Oracle-Cerner transaction, Microsoft acquired Nuance and Google partnered with MEDITECH in March 2022.

Different jurisdictions promulgate antitrust laws and regulate oversight agencies that apply policies, require good governance, impose penalties (fines). Different administrations within a jurisdiction (like the U.S.) can and do exercise their rights to pass new laws or to interpret antitrust laws differently (often by Executive Action, EO) {White & Case Insight, June 30, 2026; AI Watch: Global Regulatory Tracker – United States].

Regulatory Workarounds

Firms use a number of tactics to avoid regulatory scrutiny, including: 1) Roll-up strategy of small non-reportable buyouts to gradually build a monopoly, 2) Use of non-traditional metrics that are harder to capture in digital market ecosystems (i.e., real consumer lock-in, data dominance), and 3) Efforts to find gaps between international regulatory regimes.

Today regulators struggle to address market complexities: 1) Asymmetrical information (i.e., internal data fragmentation, proprietary algorithms, and complex networks inside multi-billion-dollar corporations), 2) High legal evidentiary thresholds to prove harms to innovation, environment, jobs, or future competition, 3) Financial lobbying to influence rules, create exemptions, and dilute enforcement resources, and 4) Hesitations to sanction “too big to fail” employers (risking job losses, market instability).

EHR Sector Oversight

Regulatory oversight of EHRs (see Table 2) includes interoperability mandates by the U.S. Office of the HHS ONC’s Assistant Secretary for Technology Policy, with regulations legally requiring that major vendors use application programming interfaces (APIs) to prevent information blocking as an anticompetitive moat. In 2024, HHS ONC set certification standards for interoperability and data sharing (extending The Cures Act HTI-1, 2023). 

Epic’s Care Everywhere is an interoperability platform for exchanging patient information across different medical groups (per HTI-1 Final Rule, 2025). While supporting shifts towards expanded data classes (per USCDI version 3), the Care Everywhere platform also adds algorithmic transparency for the use of AI tools in the Epic EHR (Epic Care Everywhere Supported Standards, June 6, 2025).

AI – EHR Business Interface - Collaboration, or Collusion?

Cloud services are buying EHR vendors. EHR vendors are investing in AI capabilities. Their goal is to develop predictive models that assist with real-time clinical decision-making (triaging, risking, POC testing). Because direct AI query of EHRs is generally unsuccessful due to incompatible data structures and intrinsic data errors, EHR vendors have had to develop proprietary highly curated databases to train and test their AI model capabilities.  

Examples of current EHR-AI connectivity include:

1. Health systems “contributing” to Epic’s COSMOS database of >300M unique patients to help Epic train it’s CURIOSITY in-hospital event prediction model [Read Here].
2. Ambient AI clinical assistants (i.e., Nuance’s DAX CoPilot, Nvidia partner’s Abridge, Suki, CarePilot, etc.) are widely used. Per a 2026 AMA survey, 81% of U.S. physicians now use AI in a professional setting, up from 38% in 2023, with 28% using AI for chart summaries or billing documentation (Read Here). Of note, Gen AI’s medical applications are not subject to current AI regulatory oversight rules (see Table 2).
3. Hospitals using AI for revenue cycle and supply chain management acquire such capabilities from EHR vendors or develop them “in house” in partnership with AI management platforms like Signal-1 to centralize governance and track ROI [Read Here].
4. Med-tech companies seeking health system data for R&D on new medical devices and/or AI-enhanced device capabilities (to be sold back into the health care sector).

To build trust, health professionals can be represented on EHR firm advisory committees like Epic’s Specialty Steering Boards. While seemingly inclusive, true database transparency remains lacking. Neither patients nor providers know what elements of the EHR are subject to AI analytics for the purposes of in-house model training and deployment feasibility testing during real-world scaling.

Serious risks to patients remain in this early AI health care era. Major health system EHR data breaches persist (March 2026 Healthcare Data Breach Report ). And leading EHR companies have weighed in on proposed federal legislation to create a dystopian national databank threatening the personal privacy protection-program integrity balance (Epic Announcement).

Comedy of the Absurd?

In the final scene of Act II, Vladimir asks, “Well, shall we go?" and Estragon replies, “Yes, let’s go."  As it was at the end of Act I, Samuel Beckett’s stage direction is…

“They do not move.” 

While governments and economic unions work diligently to legislate and regulate the involved business sectors (see Table 2), to prevent and break up monopolistic cartels, and to hold the most flagrant violators publicly accountable, powerful stock market traders and private venture capitalists have their own existential motives to oppose controls. Humanity’s fervent wish for answers (accountability for wrongs) —and our existential expectation of fairness (access for all) — remain unfulfilled by the intertwined AI tech and EHR sectors. This stalemate of uncertainty offers clear proof of how continuing never events can adversely affect the human condition.

But like the two tramps doggedly waiting for Godot, any expectation of real movement towards markets where key factors of production become more competitive (open) and better regulated (trusted) is comedy of the absurd.

They do not move.

Author 

— by Doug Miller, MD, Professor, Department of Medicine: Cardiology, Department of Artificial Intelligence & Health, Department of Radiology and Imaging, Medical College of Georgia, 9/2026

Email Your Comments!

View Blog Index