On April 7, 2025, the Centers for Medicare & Medicaid Services (CMS) released the Calendar Year (CY) 2026 Rate Announcement and final rule for Medicare Advantage (MA) and Part D programs. Notably absent from this final regulation were specific guardrails for the use of artificial intelligence (AI) that had been included in the proposed rule issued in late 2024. CMS had initially proposed these rules aiming to “ensure services are provided equitably irrespective of delivery method or origin, whether from human or automated systems.” The agency had also sought to “clarify that in the event that an MA plan uses AI or automated systems, it must … provide equitable access to services and not discriminate on the basis of any factor that is related to the enrollee’s health status.” However, in the final rule, CMS decided not to implement such proposals.
THE PREVIOUSLY ARTICULATED NEED FOR GUARDRAILS
The initial push for AI guardrails stemmed from significant concerns outlined in the proposed rule. CMS noted the “growing use of AI within the healthcare sector” and stated, “we believe it is necessary to ensure that the use of AI does not result in inequitable treatment, bias, or both… and instead is used to promote equitable access to care.”
The agency highlighted potential dangers, citing “many instances of algorithmic discrimination, where the use of AI has resulted in deepening bias and discrimination, exacerbating existing inequities.” Furthermore, CMS pointed out that “AI and related tools rely on large data sets which can have missing or incorrect information,” which “amplifies bias and may result in low quality AI recommendations.” Concerns were also directly communicated to the agency, as “CMS has received concerns from external stakeholders… about beneficiary harm potentially resulting from MA organizations’ use of AI.”
Despite these initial justifications, CMS did not provide a specific reason in the final rule for omitting the guardrails, merely stating they would not be finalized at this time. However, CMS did “acknowledge the broad interest in regulation of AI and will continue to consider the extent to which it may be appropriate to engage in future rulemaking in this area.”
REQUIREMENTS FROM PROPOSED RULE
The proposed, but now discarded, AI guardrails would have directly impacted how MA organizations deploy AI tools. The proposal would have required “MA organizations to ensure services are provided equitably irrespective of delivery method or origin, whether from human or automated systems.” This principle was intended to apply broadly, “regardless of the tools or methods utilized to make care decisions or to provide that care.” The core requirement was that “artificial intelligence or automated systems, if utilized, must be used in a manner that preserves equitable access to MA services.”
AFFECTED ENTITIES
The decision not to implement AI guardrails directly impacts several groups. Medicare Advantage Organizations (MAOs) will not face these specific federal requirements for their AI usage in the immediate future, potentially reducing administrative or compliance burdens for now. Medicare beneficiaries, particularly those from populations historically subject to health disparities or biases, will not have the specific protections these guardrails were intended to provide against potential AI-driven inequities in care access or utilization management. Health technology companies developing AI tools for healthcare may also see this as a signal regarding the current regulatory landscape. Patient advocates and policymakers who raised concerns about AI bias will likely continue to monitor CMS’s actions and advocate for future regulation.
TIMING ISSUES
The final rule containing this decision carries an effective date of June 3, 2025. The policies outlined within it, including the absence of the AI guardrails, “are applicable to coverage beginning January 1, 2026, except as otherwise noted.”
Categories: AI, Uncategorized
Leave a Reply