The Central Observation

On March 19, 2025, the Journal of the American Veterinary Medical Association published a joint position statement on artificial intelligence by the American College of Veterinary Radiology and the European College of Veterinary Diagnostic Imaging. The statement, authored by Ryan B. Appleby of the Ontario Veterinary College and four colleagues from Kansas State University, the University of Guelph, and Cornell University, established the position of the field’s specialty college on the commercial AI radiology products being marketed to veterinary clinics. The statement, in its formal published language, articulated the standards such products must meet to be safely used in clinical practice: adherence to good machine learning practices, transparency about training data and methodology, error reporting infrastructure, postimplementation monitoring, peer-reviewed third-party validation, and the maintenance of a qualified veterinarian — preferably a board-certified radiologist — in the diagnostic interpretation loop. The statement did not directly name any commercial vendor as failing to meet the standards. It did not need to. The standards themselves, applied against the commercial products visible in the U.S. veterinary market at the time of publication, identified the entire product category as not meeting the documented threshold for clinical use.

The position statement was, by every conventional measure of professional governance in U.S. veterinary medicine, the most consequential single act of professional speech on commercial AI radiology that the specialty had produced. It was peer-reviewed. It was published in the AVMA’s flagship journal. It was authored by a working group spanning the leadership of two specialty colleges across two continents. It was open-access, with the explicit institutional intention that the document be available to every veterinarian, every state regulator, every clinic owner, and every AI vendor making commercial claims about their products. If the position statement had a thesis, the thesis was that commercial veterinary AI radiology is currently a category of products being sold to clinics under marketing claims that the documented validation evidence does not support, and that the profession’s responsibility — as articulated by its specialty college — was to recognize this gap and act to close it.

Eighteen months have passed since the position statement’s publication. The article you are reading now documents what has happened in those eighteen months at each of the three institutions in U.S. companion-animal veterinary medicine that were positioned to act on the position statement’s findings: the American College of Veterinary Radiology itself, the American Veterinary Medical Association, and the American Animal Hospital Association. At each institution, the institutional capacity to act existed. At each institution, the documented professional notice that action was warranted existed. At each institution, the institutional vehicles through which action would have been formalized — consensus statements, House of Delegates resolutions, accreditation standards — were available and operational during the relevant period. At each institution, no such action was taken.

The cumulative pattern of inaction across three institutions is the central observation this article makes. It is descriptive of the documented record. The article’s analytical contribution is to identify the structural conditions under which the inaction is explicable: at all three institutions, the corporate parents of the commercial AI vendors whose products the position statement implicitly criticizes are simultaneously the financial sponsors, conference partners, and Preferred Business Providers whose continued financial relationship with the institution constitutes a substantial fraction of the institution’s operational continuity. The structural conflict of interest the relationship creates does not require any individual at any institution to have acted in bad faith for the institutional inaction to be its predictable consequence. The structure produces the result. The inaction is the result the structure produces.

What this article does not do is allege that any individual at the ACVR, the AVMA, or AAHA has been corrupted, has acted improperly, or has made any specific decision influenced by financial conflict. The article makes no such allegation, expressly disclaims any such allegation, and does not require any such allegation to make the structural argument it makes. The institutional decisions documented in this article are institutional. The corporate sponsorship architecture documented in this article is the publicly published architecture as it exists on the institutions’ own websites. The structural argument is the well-recognized regulatory capture dynamic that has been documented in academic literature on professional self-regulation in healthcare and other industries for decades. The article applies that framework, descriptively, to the documented institutional behavior in the eighteen-month window following the position statement’s publication. The reader can draw their own conclusions about what the structural pattern means for the integrity of professional self-regulation in U.S. veterinary medicine.

What the Position Statement Said, In the Authors’ Own Published Words

To establish what the institutions were on notice of, the article begins with the position statement’s own published language. Anyone reading this article should be able to verify the position statement directly: it is open-access in JAVMA, with the citation Appleby RB, DiFazio M, Cassel N, Hennessey R, Basran PS. American College of Veterinary Radiology and European College of Veterinary Diagnostic Imaging position statement on artificial intelligence. JAVMA. 2025;263(6):773-776, doi:10.2460/javma.25.01.0027. The statement is also linked from the ACVR’s own AI policy page at acvr.org/artificial-intelligence-in-veterinary-diagnostic-imaging-and-radiation-oncology, which is itself a useful additional source for the College’s broader public communication on the topic.

The position statement opens by acknowledging the transformative potential of AI in veterinary diagnostic imaging and radiation oncology, then transitions immediately to the operational requirements such systems must meet to be safely deployed in clinical practice. The published statement requires that AI systems “must adhere to good machine learning practices, emphasizing transparency, error reporting, and the involvement of clinical experts throughout development.” It requires that “these tools should also include robust mechanisms for secure patient data handling and postimplementation monitoring.” It identifies “the critical importance of maintaining a veterinarian in the loop, preferably a board-certified radiologist or radiation oncologist, to interpret AI outputs and safeguard diagnostic quality.” It calls for “rigorous peer-reviewed research, unbiased third-party evaluations, and interdisciplinary collaboration to establish evidence-based benchmarks for AI applications.” It calls for “regulatory bodies” to “establish guidelines to prevent misuse and protect the profession and patients.” It explicitly calls for veterinarians to “disclose AI usage to pet owners and provide alternative diagnostic options as needed.”

The position statement’s broader framing on the ACVR’s AI policy page — published as a companion to the formal JAVMA statement — articulates the position even more directly. The ACVR and ECVDI, per that page, “support the development and use of ethical and transparent AI in veterinary diagnostic imaging and radiation oncology applications” while simultaneously stating that “veterinarians should exercise caution when using AI in diagnostic imaging and must understand the limitations of the systems they are using.” The page additionally observes that “the legal responsibility of decisions made from any AI system has yet to be determined but is likely to have some degree of responsibility for veterinarians themselves rather than developers of the AI alone.” The framing constructs a category of products that the College is on record as believing requires caution to use, that has documented limitations the user must understand, and that may transfer legal liability for diagnostic errors to the user rather than the vendor. The College has, in effect, told its own Diplomates and the broader profession that the products currently on the market carry clinical and legal risks that a careful practitioner should recognize before using them.

Reading the position statement against the commercial products available in the U.S. veterinary market in 2025 produces an unavoidable conclusion. SignalPET’s marketing claim of “over 2 million annotated veterinary radiographs” does not include the documentation of training-data methodology that the position statement requires. Vetology’s claim of “over 300,000 Board Certified veterinary radiologist-reviewed cases” does not include the breakdown by annotation type, by annotator credential, or by independent annotator count that the position statement implicitly requires. Antech RapidRead’s claim of “16 million images sourced from an unprecedented library of more than 8 billion images” does not include the per-indication validation statistics, the algorithm version traceability, or the third-party evaluation against pathology-confirmed reference standards that the position statement identifies as necessary. The Joslyn et al. peer-reviewed commentary published in Frontiers in Veterinary Science in June 2025 — three months after the position statement’s publication — documented that the commercial veterinary AI software it evaluated was “continuously updated and does not have version numbers,” precisely the version-traceability failure mode the position statement implicitly identifies. The Ma et al. JAVMA pilot study published in 2026 documented sensitivity ranging from 71% to 90% on pathology-confirmed canine abdominal cases across multiple commercial veterinary AI services, with documented “deficiencies in interpretation.”

The position statement, applied to the actual commercial product category, identifies that category as not meeting the standards. This is not a paraphrase or interpretation that requires editorial judgment. It is the categorical implication of the published standards applied against the documented product evidence. The institutions on notice of this implication included the ACVR (which authored the statement), the AVMA (which published it in its flagship journal), and AAHA (which is institutionally responsible for the accreditation framework that defines the operational standards for AAHA-accredited veterinary practices in the United States and Canada). All three institutions have published responses on AI policy in the eighteen-month period following the publication. None of those responses includes any institutional action that would constrain the commercial product category that fails the position statement’s standards.

The ACVR’s Institutional Response to Its Own Position Statement

The American College of Veterinary Radiology is the AVMA-recognized specialty organization for veterinary radiology in the United States. Founded in 1961, the College is the credentialing authority for board-certified veterinary radiologists (DACVR) in North America, with approximately 600 active diagnostic-imaging Diplomates as of 2025. The College’s stated mission is to advance excellence in patient care through leadership, innovation, and education in veterinary diagnostic imaging, equine diagnostic imaging, and radiation oncology. Its primary institutional vehicles are the Annual Scientific Conference (held each October), continuing education programming, residency program oversight, board-certification examinations, and consensus-statement publication. The position statement on AI was the College’s most consequential institutional speech act on commercial AI radiology in the College’s history, and the closest the field’s specialty college has come to a formal institutional finding that an entire product category fails the validation standards required for clinical use.

The institutional response that has followed is documented in the College’s own publicly accessible communications, primarily through its conference programming, its Partners/Affiliates list, and its institutional structure. The 2025 Annual Scientific Conference, held October 15-18 at the Omni Shoreham in Washington, DC, was the first ACVR conference following the position statement’s publication and the natural institutional vehicle for operationalizing the statement’s findings into specific guidance, vendor evaluation criteria, or programmatic constraint on the products implicated. The conference was instead organized around the same corporate partnership architecture the College had operated for years, with the explicit institutional acknowledgment in the conference prospectus that “the support of our corporate partners plays a crucial role in advancing ACVR’s mission to promote excellence in patient care.” The corporate partners supporting the 2025 conference included, prominently, the very vendors whose products the position statement implicates.

The ACVR Partners/Affiliates List, Examined

The ACVR’s official Partners/Affiliates page at acvr.org/partners-affiliates lists the corporate entities that have ongoing partnership relationships with the College. The list is not a hidden document; it is the institution’s own published acknowledgment of the corporate relationships that fund its operational continuity. Three of the listed Partners/Affiliates are commercial veterinary AI radiology vendors whose products fall directly within the category the position statement identifies as failing validation standards.

Antech Imaging Services is a Mars Petcare subsidiary, the corporate sibling of Antech Diagnostics and the operating entity of the RapidRead AI product. Antech Imaging Services is described, in its own ACVR Partners/Affiliates listing, as committed to “providing a Better World for Pets” — a phrase identical to Mars Petcare’s corporate mission statement and a clear identification of the entity’s parent corporate structure. The Mars Petcare ownership structure encompasses Banfield Pet Hospital (acquired 2007), BluePearl Veterinary Partners (acquired 2015), VCA Inc. (acquired 2017 in a $9.1 billion all-cash transaction including $1.4 billion in outstanding debt), and Antech Diagnostics (acquired as part of the VCA transaction). The RapidRead AI product, marketed by Antech Diagnostics under the broader Mars Petcare Science & Diagnostics division, is the product whose 16-million-image training corpus claim is directly addressed in the Phantom Radiologists series that this publication has previously published.

Vetology, also listed on the ACVR Partners/Affiliates page, is described in its own listing as a company that “offers practical solutions for veterinary clinics and radiologists” with a platform combining “AI radiology reports, teleradiology services, and an AI-assisted prelim report writing platform.” Vetology is the operating entity of the Vetology AI radiology product whose marketing claim of “over 300,000 Board Certified veterinary radiologist-reviewed cases” is directly addressed in the Phantom Radiologists series.

Radimal, also listed, is described in its own listing as “the Doctor + AI solution for Veterinary Medicine.” Radimal is a smaller commercial veterinary AI radiology vendor operating in the same product category.

The structural pattern is documented on the College’s own website. The same College that authored the position statement establishing that no commercial veterinary AI radiology product meets the documented standards for safe clinical deployment is the same College whose official Partners/Affiliates list includes the operating entities of three of the commercial veterinary AI radiology products in the U.S. market. The Mars Petcare subsidiary specifically operating the RapidRead AI product whose validation deficiencies are most thoroughly documented in the broader investigative literature is one of the listed Partners. The listing exists as a matter of public record, and has not been removed or modified in the eighteen months following the position statement’s publication.

Antech’s own marketing materials confirm the active conference partnership relationship for the 2025 conference. The Antech Diagnostics page promoting attendance at the 2025 ACVR Annual Scientific Conference described the conference participation as joining “a global team of over 140 Antech radiologists and additional veterinary specialists” at the event. Antech’s booth was at positions 308/310 in the exhibit hall, with the company offering branded plushies to conference attendees and dedicated time to “talk diagnostics solutions for your practice.” The same company whose RapidRead product the position statement’s standards implicate had a corporate booth at the same conference where the position statement’s authors and ACVR leadership would have been present, with no apparent institutional friction between the position statement’s findings and the corporate partnership relationship.

The Joint AVMA-ECVDI Position Statement Was Not Followed By a Vendor Evaluation Framework

One of the structural questions a position statement of this kind raises, in any properly functioning regulatory ecosystem, is what mechanism the issuing institution will use to translate the statement’s findings into operational consequence for the regulated entities. In human medicine, when a specialty college issues a position statement establishing that a category of products does not meet required standards, the natural institutional follow-through includes vendor evaluation programs, product certification frameworks, clinical practice guidelines incorporating the standards, and continuing education programming designed to translate the standards into clinical decision-making by the College’s own Diplomates. The American College of Cardiology, the American College of Radiology, and the American Society of Clinical Oncology each operate institutional vehicles of this kind for the categories of medical devices, drugs, and clinical services within their respective specialty domains.

The ACVR has not established such a mechanism for commercial veterinary AI radiology in the eighteen months following the position statement’s publication. The College has not published a list of commercial AI products that meet the position statement’s standards. The College has not published a list of commercial AI products that fail the position statement’s standards. The College has not established a third-party evaluation framework against which products could be assessed. The College has not modified its conference partnership criteria to require that participating AI vendors produce the disclosure documentation the position statement implicitly requires. The College has not issued additional consensus statements operationalizing the position statement’s standards into specific clinical guidance for Diplomates evaluating AI products in their own practice or referring clinicians’ practices.

The institutional capacity to do any of these things has existed throughout the eighteen-month period. The decision not to do them is itself the institutional response, made meaningful by the absence of any alternative response that would have constrained the products at issue. Position statements without operational follow-through function, in the language of regulatory analysis, as expressions of professional preference rather than as constraints on commercial behavior. The ACVR’s position statement on AI, in the absence of operational follow-through, is currently functioning in the former category. The reason for the absence of operational follow-through is the question the structural argument in this article addresses, with reference to the corporate sponsorship architecture documented above.

The 2026 Conference Has Continued the Same Architecture

The 2026 ACVR Annual Scientific Conference is scheduled for New Orleans in October 2026. The conference prospectus, as of the time of this article’s preparation, has continued the same corporate partnership architecture as the 2025 conference. The Partners/Affiliates list as published on the ACVR website continues to include Antech Imaging Services, Vetology, and Radimal as official partners. The College has not announced any modification to its corporate-partner evaluation criteria in the period leading up to the 2026 conference. The institutional pattern that the 2025 conference established — that the position statement’s findings did not translate into modification of the College’s corporate partnership architecture — has been continued into the 2026 conference cycle.

The AVMA’s Institutional Architecture and the Mars Petcare Relationship

The American Veterinary Medical Association is the largest membership organization for veterinarians in the United States, with more than 100,000 member veterinarians as of 2025. Founded in 1863, the AVMA is the profession’s primary representative body in legislative advocacy, standards development, accreditation of veterinary medical education programs (through the AVMA Council on Education), publication of the profession’s flagship journal (JAVMA), and operation of the AVMA Convention each summer. The AVMA’s House of Delegates is the policy-making body of the organization, with elected delegates from each state veterinary medical association and from the major specialty colleges. Policy resolutions advanced through the House of Delegates represent the AVMA’s official institutional position on matters of professional concern.

The AVMA’s most consequential single act on the position statement was its publication. The position statement appeared in JAVMA, the AVMA’s flagship journal, on March 19, 2025, as a peer-reviewed open-access publication. The editorial decision to publish the statement was itself an institutional act of substantial weight: it gave the position statement the formal authority of peer-reviewed publication in the profession’s most authoritative journal, and made the statement’s findings available to every AVMA member, every state veterinary medical board, and every veterinary practice in the United States. The publication, by itself, was the AVMA’s most consequential institutional response to the standards the position statement articulated.

What did not follow, in the eighteen months since publication, is institutional response of any kind that would translate the position statement’s findings into operational policy. The AVMA House of Delegates met in summer 2025 and again in winter 2025-2026, with the standard resolution-deliberation calendar that produces formal AVMA policy. No resolution was advanced during these meetings that operationalized the position statement’s findings into AVMA policy. No amendment was advanced to the AVMA Model Veterinary Practice Act language addressing AI-assisted diagnostic interpretation. No modification was made to the AVMA’s existing corporate partnership framework to address vendors whose products fail the position statement’s validation standards. No formal communication was issued from the AVMA’s Council on Veterinary Service or from the AVMA’s existing committee structure addressing the implications of the position statement for member veterinarians evaluating AI products for adoption in their own practices.

The absence of institutional response is consequential because the AVMA, more than any other institution in U.S. companion-animal veterinary medicine, has the political and policy reach to translate professional consensus into operational standards. AVMA policy resolutions are routinely cited by state veterinary medical boards in the development of state-level practice standards. AVMA Model Practice Act language is the foundation document on which most state veterinary practice acts are based. AVMA accreditation through the Council on Education is the credentialing mechanism for the U.S. and Canadian veterinary medical schools. The AVMA’s institutional capacity to give operational consequence to the position statement’s findings is, in any practical sense, the largest institutional capacity for such action in the U.S. companion-animal veterinary medicine ecosystem. The capacity has not been used.

The Documented Mars Petcare-AVMA Operational Partnership

The Mars Petcare-AVMA institutional relationship extends substantially beyond passive corporate sponsorship of conference programming or advertising in JAVMA. The relationship includes operational program partnerships, jointly developed published resources, and visible cooperation across multiple programs that suggest a degree of institutional coordination beyond what would be expected of a typical regulated-entity relationship.

The most prominently documented such collaboration is the Positive Pet Care Guide initiative, announced in September 2022 at the 21st annual Banfield Pet Healthcare Industry Summit. The initiative was a working group convened by Mars Veterinary Health (the Mars Petcare division comprising Banfield, BluePearl, VCA, and other practice networks) with the AVMA and several other industry organizations including the American Association of Veterinary Medical Colleges, the National Association of Veterinary Technicians in America, Not One More Vet, the Veterinary Hope Foundation, and Veterinary Medical Association Executives. The working group’s stated objective was to “create a list of shared expectations and responsibilities that the profession and pet owners alike can leverage to further a mutual understanding and respect towards one another.” The Positive Pet Care Guide was published in 2023 as the working group’s output and became a jointly branded resource available to the broader profession.

The Banfield Pet Healthcare Industry Summit itself, hosted annually by Mars Veterinary Health, has featured AVMA leadership as panelists and speakers across multiple consecutive years. The 2022 Summit panel on mental health and professional issues included AVMA President Lori Teller alongside speakers from Mars Veterinary Health subsidiaries BluePearl and VETgirl. The 2024 Summit announced Mars Veterinary Health as the founding sponsor of the Veterinary Sustainability Alliance, with continued AVMA institutional participation. The Summit programming consistently positions the AVMA as a partner organization rather than as a regulator with which Mars Veterinary Health interacts at arm’s length.

Banfield Pet Hospital — the Mars Petcare subsidiary — additionally invests in expanding the AVMA’s member toolkit for managing hospital online reputations, with the resulting resources made available to all veterinary professionals at no cost. This is not a one-time sponsorship; it is ongoing operational support for an AVMA member benefit program. The cumulative pattern is one of institutional cooperation, jointly developed resources, and shared programmatic objectives. None of this is improper in itself. It is, however, the documented institutional context within which the AVMA’s response (or absence of response) to the position statement’s findings should be read.

The AVMA’s 2017 Editorial Treatment of the Mars-VCA Acquisition

An additional piece of institutional context that bears on the AVMA’s relationship with Mars Petcare is the AVMA’s editorial treatment of the Mars-VCA acquisition in JAVMA News at the time of the transaction in 2017. The JAVMA News article “Veterinarians Incorporated” published March 1, 2017, documented the $9.1 billion all-cash acquisition of VCA Inc. by Mars and the resulting consolidation of Banfield, BluePearl, VCA, Pet Partners, and Antech Diagnostics under the Mars Petcare corporate structure. The article quoted Dr. Scott Spaulding, a member of the AVMA Veterinary Economics Strategy Committee, characterizing the merger as “a potential positive for the veterinary industry” and describing the $9.1 billion of capital “coming into the veterinary industry” as “definitely needed.” The article reported that “the AVMA, for its part, says it understands that there are those who are supportive and those who are concerned,” and quoted the AVMA’s institutional position as being to “support every veterinarian engaged in veterinary medicine, no matter where they practice.”

The AVMA’s editorial framing of the Mars-VCA acquisition was institutionally neutral verging on supportive, with the only critical voices in the article coming from external sources — a Bloomberg Businessweek article that featured a veterinarian accusing Banfield of “pushing veterinarians to put profit ahead of the health and safety of animals.” The AVMA’s own institutional speech around the acquisition did not include the structural concerns about corporate consolidation that other professional bodies in similar regulated industries have voiced when comparable consolidations have occurred. The pattern of institutional framing that began with the AVMA’s 2017 treatment of the Mars-VCA acquisition has continued through the 2022-2025 period of joint programmatic partnership documented above.

Why the AVMA’s Institutional Inaction Matters Differently Than the ACVR’s

The ACVR’s institutional inaction in the eighteen months following the position statement’s publication is meaningful because the ACVR is the specialty college that authored the statement and is the credentialing authority for the specialty whose standards the statement articulates. The AVMA’s institutional inaction is meaningful for a different reason: the AVMA is the institution with the broadest political and policy reach to translate professional consensus into operational standards across the U.S. veterinary medicine landscape.

If the AVMA House of Delegates had advanced a policy resolution operationalizing the position statement’s findings — for example, a resolution recognizing the ACVR/ECVDI standards as the operational threshold for safe clinical deployment of commercial AI radiology products in U.S. veterinary practice — the resolution would have carried substantial force. State veterinary medical boards routinely cite AVMA policy resolutions in their development of state-level practice standards. The AVMA Model Veterinary Practice Act, which provides the foundation language for most state veterinary practice acts, could have been amended to address AI-assisted diagnostic interpretation in language consistent with the position statement’s standards. Either of these institutional acts, if undertaken, would have begun the process of translating the position statement’s findings into operational practice constraint at the state regulatory level.

Neither has occurred. The institutional capacity to undertake either has existed throughout the relevant period. The institutional silence on operationalizing the position statement’s findings is the meaningful institutional response, made meaningful by the absence of any alternative response. The structural conflict of interest between the AVMA’s institutional partnerships with Mars Petcare entities and the operationalization of standards that would constrain Mars Petcare’s commercial AI radiology product is the structural condition under which the silence is most parsimoniously explicable.

What AAHA Is, and Why Its Inaction Is Different

The American Animal Hospital Association is the only voluntary accrediting body for companion-animal veterinary hospitals in the United States and Canada. Founded in 1933, headquartered in Lakewood, Colorado, AAHA accredits approximately 3,800 to 4,000 veterinary practices currently, representing roughly 12 to 15 percent of the U.S. companion-animal practice population. AAHA accreditation is the closest equivalent veterinary medicine has to the institutional accreditation that the Joint Commission provides for human hospitals. Practices that hold AAHA accreditation publicly market the credential to clients, reference it in continuing education and referral relationships, and use it as a signal of meeting elevated standards of care.

The AAHA accreditation process requires meeting nearly 900 individual standards across approximately 50 categories: patient care, anesthesia, pain management, surgical safety, dental care, infection prevention, medical records, diagnostic imaging quality, personnel credentialing, client communication, continuing education, and others. Practices undergo on-site evaluations every three years against these standards. Practices that fail to meet the standards lose accreditation. The mechanism for institutional enforcement of professional standards in companion-animal veterinary practice in the United States, to the extent any such voluntary mechanism exists, runs through AAHA more directly than through any other body.

The relevance of this institutional capacity to the position statement’s findings is direct. AAHA’s existing standards explicitly address diagnostic imaging quality, the credentialing of personnel performing diagnostic interpretation, and the requirement that diagnostic decisions affecting patient care be supported by appropriate quality assurance. These existing standards constitute the natural institutional vehicle through which an AAHA accreditation-level constraint on commercial AI radiology products that fail validation requirements could be operationalized. An amendment to the AAHA Diagnostic Imaging Standards requiring that any AI radiology product used in an accredited practice meet documented validation standards equivalent to what the ACVR position statement called for would, if implemented, immediately change the commercial calculus for the vendors. A vendor whose product is incompatible with AAHA-accredited practice loses access to the most commercially desirable veterinary clinics in North America. The vendors would face a powerful incentive to produce the validation documentation the position statement called for, because their access to AAHA-accredited practices would otherwise be at risk.

AAHA has not made this amendment. The most consequential institutional opportunity to do so, in the eighteen months following the position statement’s publication, was the comprehensive Standards of Accreditation refresh that AAHA itself characterized as the most significant institutional standards review in the organization’s 90-year history. That refresh did not include any standard addressing commercial AI radiology validation. The institutional opportunity was used for other purposes.

The September 2025 AAHA Standards Refresh — and What It Did Not Address

On September 2025, AAHA announced the completion of what the organization described as “the first comprehensive refresh of the Standards of Accreditation in the Association’s 90-year history.” The announcement, published on the AAHA website, characterized the refresh as a “top-to-bottom review” marking “a historic milestone in our mission to elevate veterinary care.” The refresh process had been ongoing for more than a year, with substantial institutional resources devoted to reviewing every standard and developing the new framework.

The first batch of new standards from the refresh was released July 22, 2025, three months before the September announcement. Per AAHA’s published description, the four new Standards of Accreditation released as the first headline outcomes of the refresh were:

A Microchipping Standard requiring that AAHA-accredited practices initiate conversations about microchipping at a pet’s first visit and verify functionality and location annually.

A Nontherapeutic Procedures Standard discouraging accredited practices from performing or referring nontherapeutic procedures when not medically necessary, including specific procedures the standard identifies that “should only be performed when a clear medical indication exists.”

Two additional headline standards addressing other clinical practice domains.

Each of these new standards is, in itself, a defensible institutional response to a documented professional concern. Microchipping policies, nontherapeutic procedure policies, and the other domains addressed are legitimate areas for accreditation-level standards development. The question this article raises is not whether the new standards are good standards. The question is what the comprehensive refresh did not include.

None of the four new standards addressed commercial veterinary AI radiology validation. None addressed the algorithm version traceability that the Joslyn et al. commentary documented as missing from the commercial AI products it evaluated. None addressed the specialist-oversight requirements the ACVR/ECVDI position statement explicitly called for. None addressed the third-party evaluation requirements the position statement identified as necessary. None addressed the disclosure-to-pet-owner requirements the position statement explicitly called for. The comprehensive Standards refresh — the institutional vehicle most directly capable of translating the position statement’s findings into accreditation-level operational constraint on the commercial product category — did not include any such constraint.

The institutional capacity to include such a standard in the comprehensive refresh existed throughout the refresh process. The ACVR/ECVDI position statement was published in March 2025, four months before the first batch of new standards was released in July 2025 and six months before the September 2025 announcement of the broader refresh. AAHA had access to the position statement, had the institutional vehicles to incorporate its standards into accreditation criteria, and had ongoing programmatic capacity to develop new standards as part of the comprehensive refresh. The decision not to include an AI accreditation standard in the comprehensive refresh was the institutional response, made meaningful by the absence of any alternative response that would have provided accreditation-level enforcement of the position statement’s findings.

The AAHA Preferred Business Provider Program and the Vendor Sponsor Architecture

The structural conditions that explain AAHA’s institutional inaction are documented in AAHA’s own published partnership architecture. The AAHA Preferred Business Provider program, accessible at aaha.org/for-veterinary-professionals/savings-programs/preferred-providers, identifies the corporate vendors that have formal partnership relationships with AAHA, with participating vendors offering “preferred pricing or other special services to AAHA members.”

The IDEXX Laboratories-AAHA Preferred Business Provider relationship was formalized in November 2012 through a strategic partnership announcement. AAHA’s own announcement at the time stated that the agreement would “offer tools and resources to AAHA members for a client-centric approach to practice management” and provide AAHA members with discounted access to the IDEXX Cornerstone Practice Management System and IDEXX Pet Health Network Pro. The relationship has continued for more than thirteen years, with periodic public renewal and program development. IDEXX operates commercial veterinary AI radiology products that fall within the category the position statement identifies as failing validation requirements.

The Antech Diagnostics-AAHA institutional relationship is similarly long-standing. Antech, as a Mars Petcare Science & Diagnostics subsidiary, operates the RapidRead AI product whose validation deficiencies are directly addressed in the Phantom Radiologists series. The Mars Petcare corporate structure encompassing Antech also operates Banfield, BluePearl, and VCA, the largest practice networks in the U.S. companion-animal veterinary medicine landscape. The Mars Petcare corporate parent has, by trade-press characterization, been described as “the biggest vet provider in the country,” with operational presence across primary care, specialty referral, diagnostic laboratory services, imaging hardware, and AI radiology software.

The structural conflict of interest the Preferred Business Provider relationships create for AAHA is direct. An AAHA accreditation standard requiring that commercial AI radiology products meet documented validation requirements would, if rigorously applied, constrain the commercial viability of the AI products operated by AAHA’s largest Preferred Business Providers. The institutional incentive to develop such a standard runs against the institutional incentive to maintain favorable Preferred Business Provider relationships with the corporate parents whose products would be constrained. The structural conflict does not require any individual at AAHA to have made any specific decision in bad faith for the institutional inaction to be the predictable consequence. The structure of the relationship produces the predictable result.

The Contrast: When AAHA Has Acted Decisively on Other Standards

An important piece of evidence that the AAHA institutional inaction on AI accreditation standards is structural rather than incidental is the contrast with AAHA’s responsiveness in other areas where the corporate-sponsor entanglement does not exist. AAHA’s institutional record on standards development in domains without significant corporate-sponsor financial exposure is one of consistent responsiveness to professional notice. The AAHA Anesthesia Safety and Monitoring Guidelines have been updated multiple times in the past decade to reflect evolving professional consensus. The AAHA Pain Management Guidelines have been the subject of multiple iterations. The AAHA Dental Care Guidelines have been updated to reflect new clinical evidence. The AAHA Nutrition Guidelines have been refreshed to address current professional understanding. In each of these cases, the institution moved with reasonable speed from professional notice to standards modification, demonstrating that the institution has the capacity and the operational mechanism to translate professional consensus into accreditation-level standards.

The AI radiology validation question presented AAHA with the same professional-notice-to-standards-modification cycle. The notice was the ACVR/ECVDI position statement. The institutional vehicle was the comprehensive Standards of Accreditation refresh. The capacity existed. The operational mechanism existed. The only feature distinguishing the AI radiology validation question from the anesthesia, pain management, dental care, and nutrition standards areas where AAHA has acted is the corporate-sponsor financial entanglement. The contrast is the evidence that the inaction is structural rather than incidental — that the institution has the capacity to act, has the operational mechanism, and has the documented professional notice, but has not acted in this specific area where the corporate-sponsor structural conflict exists.

What an AAHA Accreditation Standard on AI Would Have Looked Like

To make the institutional choice visible, it is useful to specify what an AAHA accreditation standard on AI radiology validation would have included if AAHA had elected to develop one in the September 2025 comprehensive refresh. The standard would have been operationally specific, evaluable through on-site inspection, and rigorous enough to meaningfully constrain the commercial product category that fails the position statement’s standards.

The standard might have required, for any commercial AI radiology product used in an AAHA-accredited practice, documented evidence of: training-data composition disclosed at the level of detail the CLAIM checklist (Checklist for Artificial Intelligence in Medical Imaging, Mongan, Moy, and Kahn, Radiology: Artificial Intelligence, 2020) requires; per-indication validation statistics including sensitivity, specificity, false-positive rate, and false-negative rate against a defined reference standard; algorithm version traceability with documented change-control policy; specialist oversight at the level of board-certified veterinary radiologist review of AI output for any clinical decision turning on the AI’s interpretation; and disclosure to pet owners that AI was used in the diagnostic interpretation, with availability of alternative diagnostic options as the position statement explicitly called for.

The accreditation enforcement mechanism would have required AAHA-accredited practices to maintain documentation of their AI vendor’s compliance with the standard, with the on-site evaluation including review of the documentation. Practices using AI products that did not meet the standard would have been required to discontinue the use as a condition of maintaining accreditation. The standard’s effect on the commercial AI vendor market would have been substantial: vendors whose products fail the validation requirements would have been incompatible with AAHA-accredited practice, immediately reducing their commercial market by the AAHA-accredited fraction (12-15% of U.S. clinics, but representing the most commercially desirable clinics in the practice population).

The vendors, faced with this commercial constraint, would have had a strong incentive to produce the validation documentation the standard required. The position statement’s findings would have been operationalized into accreditation-level enforcement. The professional self-regulation mechanism would have functioned the way it is institutionally supposed to function: professional consensus identifies a standard, the accrediting body operationalizes the standard into enforceable accreditation criteria, the regulated entities adapt to the standard or lose access to the accredited market.

This is what did not happen. The institutional choice to leave the comprehensive Standards refresh without any AI standard addressing the position statement’s findings is the choice that this article documents. The choice was made in the same institutional context that includes IDEXX Laboratories and Antech Diagnostics as Preferred Business Providers, and the corporate parents of those entities (IDEXX Laboratories Inc. and Mars Petcare) as the largest single financial relationships in the AAHA institutional portfolio.

The Structural Argument, Stated Plainly

The cumulative pattern across the ACVR, the AVMA, and AAHA is, on its face, three independent institutions producing the same response to the same documented professional notice. Each institution had the institutional capacity to act on the position statement’s findings. Each institution had the operational vehicles available — consensus statements, House of Delegates resolutions, accreditation standards — through which the action would have been formalized. Each institution had documented professional notice that action was warranted. Each institution faced the same corporate sponsorship architecture connecting it to the corporate parents of the AI vendors at issue. Each institution produced the same response: institutional silence on operationalizing the position statement’s findings into commercial constraint, combined with continued institutional partnership relationships with the corporate parents whose products the position statement implicates.

The probability that three independent institutions would, by coincidence, produce the same pattern of response to the same notice in the same window is low. The structural argument is that the institutions are not, in fact, fully independent on this question. The corporate sponsorship architecture connecting all three institutions to the same set of corporate parents — Mars Petcare and IDEXX Laboratories most prominently — creates a shared structural condition under which the institutional inaction is the predictable result. Each institution, faced individually with the choice of either operationalizing the position statement’s findings (and constraining the corporate parents whose financial sponsorships fund operational continuity) or maintaining the institutional silence (and preserving the financial sponsorship architecture), produced the same choice. The structural conflict of interest does not require any communication between institutions, any coordinated decision, or any individual at any institution acting in bad faith. The structure of the relationships produces the predictable result at each institution independently.

This is the regulatory capture dynamic that academic literature on professional self-regulation in healthcare and other industries has documented for decades. The dynamic is structural, not personal. It does not require corruption, malfeasance, or bad faith on the part of any individual. It requires only that institutional operational continuity depend on financial relationships with the regulated entities, and that the institution face a choice between maintaining those relationships and undertaking enforcement that would constrain the entities. Faced with this choice repeatedly across multiple decision points, institutions positioned this way reliably produce the same pattern: speech that articulates standards, combined with institutional inaction that fails to operationalize the standards into enforcement. The inaction is the predictable result.

The Comparison: How Human Medicine’s Equivalent Institutions Are Differently Situated

The contrast that puts the veterinary side’s structural problem into focus is the architecture of professional self-regulation in U.S. human medicine. The American Medical Association, the various human-medicine specialty colleges (American College of Cardiology, American College of Radiology, American Society of Clinical Oncology, etc.), and the Joint Commission for hospital accreditation each operate within a framework where their enforcement decisions are constrained, supplemented, and often pre-empted by external regulatory bodies that the veterinary side simply does not have. The Food and Drug Administration regulates medical devices, including AI/ML-enabled medical devices, with pre-market clearance requirements that constrain commercial product entry into the human-medicine market. The Centers for Medicare & Medicaid Services regulate hospital and physician billing practices, with documented audit infrastructure that constrains diagnostic decision-making. State medical boards have disciplinary authority over individual physicians, with formal due process and documented enforcement records.

The professional self-regulating institutions on the human side operate in this regulated environment as one component of a multi-institution accountability framework, not as the sole institution responsible for translating professional consensus into commercial constraint. When the American College of Cardiology issues a position statement on a category of medical devices, the operational consequence of the statement is shaped not only by the College’s institutional response but by the FDA’s regulatory framework, by CMS reimbursement decisions, by state medical board enforcement of practice standards, and by malpractice carrier review of clinical practice. The College’s institutional inaction, if it occurred, would not be the only institutional response operating on the regulated category. Other institutions would also be acting, with their own enforcement mechanisms, their own incentive structures, and their own corporate-relationship dynamics distinct from the College’s.

The veterinary side does not have these external institutional checks. The FDA does not regulate veterinary AI software (with limited exceptions for products that affect human health). CMS does not exist for veterinary medicine. State veterinary medical boards have disciplinary authority over individual veterinarians but not direct regulatory authority over commercial AI products or their vendors. Malpractice carrier review of veterinary practice is substantially less institutionally developed than the human-medicine equivalent. The professional self-regulating institutions on the veterinary side — the ACVR, the AVMA, AAHA — are, in effect, the entire accountability framework for commercial veterinary AI products. When these three institutions choose not to act on a position statement’s findings, the effect is not partial enforcement that other institutions complete; it is the absence of any enforcement at all.

This is the structural feature that makes the veterinary side’s regulatory capture dynamic operationally consequential in a way that the human side’s equivalent dynamic, when it occurs, typically is not. On the human side, regulatory capture at one institution is partially compensated by enforcement at other institutions. On the veterinary side, regulatory capture at the three professional self-regulating institutions documented in this article produces, in effect, complete absence of operational enforcement on the regulated commercial product category. The position statement, in this institutional environment, becomes professional speech without commercial consequence.

What This Means for Veterinary Practice and Patient Care

The cumulative institutional inaction documented in this article has direct consequences for veterinary practice and patient care. Commercial AI radiology products that do not meet the validation standards the field’s specialty college has identified as required continue to be marketed to veterinary clinics, sold under the implicit warranty of conventional commercial product offerings, and deployed in clinical workflows where their output influences diagnostic decisions affecting patients. Veterinary clinics evaluating these products for adoption do not have access to the institutional accountability mechanisms — accreditation standards, model practice act language, formal vendor evaluation frameworks — that would constrain the products’ commercial availability if such mechanisms existed. The clinics are, in effect, on their own to evaluate whether the products meet the standards the position statement articulates, with no institutional support for that evaluation.

The veterinarians whose names appear on AI-generated reports as the supervising specialist are similarly unprotected by the institutional accountability framework that the position statement was, in part, intended to construct. The “veterinarian in the loop” requirement the position statement articulates — that a qualified veterinarian, preferably a board-certified radiologist, must be involved in interpreting AI outputs — is currently a recommendation without institutional enforcement. Specialists working within corporate veterinary medicine ecosystems where AI subscription is part of the corporate parent’s commercial offering have no external institutional mechanism through which to insist on the supervisory standard the position statement articulates. The structural absence of audit mechanism documented in the closing of the Phantom Radiologists series — that there is no Medicare equivalent to verify that nominally specialist-reviewed reads in this corporate structure actually involved meaningful specialist review — is the operational consequence of the institutional inaction this article documents.

The pet owners whose animals are receiving diagnostic interpretation from these products are the most directly affected by the institutional inaction. The position statement explicitly called for veterinarians to “disclose AI usage to pet owners and provide alternative diagnostic options as needed.” This recommendation has not been operationalized into any AAHA accreditation standard, AVMA model practice act amendment, or ACVR consensus statement that would require the disclosure. Pet owners are, in most cases, not informed when AI is used in the diagnostic interpretation of their pet’s radiograph. The clinic does not disclose it; the institutional accreditation framework does not require disclosure; the AI vendor does not market the AI involvement to the pet owner. The pet owner is paying for diagnostic interpretation under the implicit understanding that the interpretation reflects veterinarian judgment, when in some fraction of cases the interpretation reflects AI output that the supervising veterinarian may or may not have meaningfully reviewed.

None of this is unfixable. All of it could be addressed through the institutional vehicles that the ACVR, the AVMA, and AAHA already operate. The fact that none of it has been addressed in the eighteen months following the position statement’s publication is the institutional record this article documents. The reasons for the institutional record are the structural conditions this article analyzes. The implications for the integrity of professional self-regulation in U.S. companion-animal veterinary medicine are the question this article leaves with the reader.

---

---