United States - Georgia - AI in Insurance (HB 887)

Overview

Georgia House Bill 887 (HB 887) represented a significant legislative effort in the 2023-2024 session of the Georgia General Assembly to address the burgeoning use of artificial intelligence (AI) and automated decision-making tools within sensitive sectors. Introduced on January 9, 2024, the bill aimed to establish crucial guardrails around the application of AI in insurance coverage determinations, healthcare decisions, and the allocation of public assistance. At its core, HB 887 sought to prevent decisions in these vital areas from being made solely by AI systems, emphasizing the necessity of human oversight. The proposed legislation mandated that any decision influenced by AI or automated tools must undergo a 'meaningful review' by a human individual possessing the explicit authority to override the AI's recommendations. This provision was intended to ensure accountability and prevent potential biases or errors inherent in AI systems from having unchecked impacts on individuals' lives.

Beyond its general prohibition on sole reliance, HB 887 specifically delineated responsibilities for various entities. It tasked the Georgia Composite Medical Board with the authority to promulgate rules and regulations for the implementation and enforcement of provisions related to healthcare decisions, acknowledging the specialized nature of medical practice. Similarly, it implied that insurers would need to establish their own internal procedures for meaningful human review in coverage determinations. Despite its ambitious scope and relevance to contemporary technological advancements, HB 887 ultimately did not progress through the legislative process, dying in committee during the 2023-2024 regular session. Its introduction, however, underscored a growing recognition among state legislatures of the need to proactively regulate AI, particularly in areas with direct impacts on consumer welfare and fundamental rights.

Definitions

Georgia HB 887 provided specific definitions crucial to understanding its scope and application. The bill defined "artificial intelligence" broadly as a machine-based system capable of making predictions, recommendations, or decisions that influence a real or virtual environment, based on a given set of human-defined objectives. This definition encompasses a wide array of AI technologies, from sophisticated machine learning algorithms to more basic rule-based expert systems, as long as they contribute to decision-making processes. The breadth of this definition reflects an intent to future-proof the legislation against rapidly evolving AI capabilities, ensuring that the regulatory framework could apply to new forms of AI as they emerge.

Complementing the definition of AI, the bill also introduced the term "automated decision tool." This was defined as a system or service that leverages artificial intelligence and has been specifically developed, marketed, or modified to either make or be a controlling factor in making consequential decisions. The inclusion of "consequential decisions" highlights the bill's focus on high-stakes applications of AI, where the outcomes directly affect individuals' rights, access to services, or financial well-being. By distinguishing between general AI systems and those specifically designed for consequential decision-making, the bill aimed to target regulatory efforts where human oversight was deemed most critical. The combination of these definitions established a clear perimeter for the bill's prohibitions and requirements for human review, ensuring that the legislative intent was precisely applied to relevant AI applications.

Governance and Institutional Framework

The proposed governance structure within Georgia HB 887 placed significant responsibility on existing state regulatory bodies, particularly the Georgia Composite Medical Board. For healthcare-related decisions influenced by AI, the bill explicitly granted the Medical Board the authority and mandate to adopt and promulgate rules and regulations. This delegation of authority was critical, as it recognized the specialized expertise required to oversee AI implementation within the complex and sensitive healthcare sector. The Medical Board would have been responsible for establishing standards necessary to implement the Code section, including, but not limited to, procedures for disciplining physicians who failed to comply with the bill's requirements. This approach aimed to integrate AI oversight into established professional regulatory frameworks, leveraging existing enforcement mechanisms and institutional knowledge.

For the insurance sector, while the bill did not explicitly name a specific regulatory body for rulemaking akin to the Medical Board, it implicitly placed the onus on individual insurers. The legislation stipulated that any AI-influenced coverage determination must be meaningfully reviewed "in accordance with procedures for such review created by each insurer." This implies that each insurance company would be responsible for developing and implementing its own internal governance framework and protocols to ensure compliance with the human review mandate. This decentralized approach for the insurance industry would have required insurers to integrate AI governance into their existing compliance and risk management structures, potentially leading to a diverse set of internal policies across the industry, all aimed at achieving the bill's core objective of human oversight.

Key Focus Areas

Georgia HB 887 concentrated its regulatory efforts on three critical sectors: insurance coverage, healthcare, and public assistance. In each of these areas, the central tenet of the bill was the prohibition against making decisions based solely on the output of artificial intelligence or automated decision tools. This core restriction aimed to prevent fully autonomous AI systems from dictating outcomes that have profound impacts on individuals' lives, ensuring that a human element remains in the decision-making loop. For insurance coverage, this meant that AI could assist in processing claims or underwriting, but the final determination of whether to award, deny, reduce, or terminate coverage could not rest solely on an algorithm's recommendation.

Crucially, the bill mandated a "meaningful review" by an individual with the authority to override the AI's recommendations. This requirement was designed to ensure that human oversight was not merely a rubber-stamping process but an active and substantive evaluation. In healthcare, this applied to any decision-making process concerning care, treatment, service, or procedure, prohibiting actions based solely on AI results and requiring human review with override authority. Similarly, for public assistance, decisions regarding the award, denial, reduction, or termination of aid could not be made without this human intervention. This consistent requirement across all three sectors underscored the legislature's intent to safeguard individuals from potentially adverse or biased AI-driven decisions in areas fundamental to their well-being and access to essential services.

Implementation Framework

The implementation framework envisioned by Georgia HB 887 was multi-faceted, relying on both direct legislative mandates and delegated rulemaking authority. For the healthcare sector, the bill explicitly tasked the Georgia Composite Medical Board with the responsibility to "adopt and shall have the authority to promulgate rules and regulations governing and establishing the standards necessary to implement this Code section." This delegation was crucial, as it allowed a body with specialized expertise to develop detailed guidelines that would ensure the practical and effective integration of human oversight into AI-assisted healthcare decisions. Such rules would likely cover aspects like the qualifications of individuals conducting the review, the documentation required for AI-influenced decisions, and the specific processes for overriding AI recommendations, thereby providing clarity for healthcare providers.

In the insurance domain, the bill stipulated that any AI-influenced coverage determination must be subjected to a meaningful review "in accordance with procedures for such review created by each insurer." This placed the primary responsibility for developing implementation procedures directly on insurance companies. Each insurer would be required to establish internal protocols, training programs, and accountability measures to ensure that their use of AI in coverage decisions adhered to the bill's mandate for human oversight and override capability. While this approach offered flexibility to insurers in tailoring procedures to their specific operations, it also implied a need for robust internal compliance mechanisms. For public assistance, similar procedures for meaningful review were to be created by the relevant board, ensuring a consistent approach to human oversight across all regulated sectors. The overall framework aimed to embed human accountability at various levels of AI deployment in critical public-facing services.

Monitoring and Evaluation

While Georgia HB 887 did not explicitly detail a comprehensive monitoring and evaluation framework, its provisions implied mechanisms through which compliance and the effectiveness of its mandates would be assessed. For healthcare, the delegation of rulemaking authority to the Georgia Composite Medical Board suggested that the Board would naturally incorporate monitoring and evaluation components into its promulgated regulations. This could include requirements for healthcare providers to report on their use of AI, maintain records of human reviews and overrides, and demonstrate adherence to established standards. The Board, as the disciplinary body for physicians, would also be responsible for investigating complaints related to AI use and enforcing compliance, thereby acting as a primary monitoring entity.

In the insurance sector, the requirement for "procedures for such review created by each insurer" implicitly placed the burden of internal monitoring and evaluation on individual insurance companies. Insurers would likely need to establish internal audit functions to ensure that their AI-driven decision-making processes, including the human review component, were functioning as intended and in compliance with the law. External oversight could potentially come from the Georgia Department of Insurance, which regulates the insurance industry and would likely be responsible for ensuring that insurers meet their statutory obligations, including those related to AI use. For public assistance, similar oversight would fall to the relevant state agencies or boards responsible for administering these programs. The effectiveness of the bill, had it been enacted, would therefore largely depend on the robustness of these delegated and implicit monitoring mechanisms, coupled with the potential for consumer complaints to trigger investigations and enforcement actions.

Penalties, Liability, and Appeals

Georgia HB 887 included provisions that would have established grounds for penalties and disciplinary actions, particularly within the healthcare sector. The bill explicitly stated that the Georgia Composite Medical Board would have the authority to promulgate rules and regulations, "specifically including but not limited to the disciplining of a physician who fails to comply with this Code section." This provision clearly outlined a mechanism for holding healthcare professionals accountable for non-compliance with the mandated human review and override requirements when using AI in decision-making. The nature of these disciplinary actions could range from warnings and fines to license suspension or revocation, depending on the severity and frequency of violations, as determined by the Medical Board's future rulemaking.

While the bill did not explicitly detail specific penalties for insurers or public assistance agencies for non-compliance, the general framework of regulatory oversight implies that existing legal and administrative remedies would apply. For insurers, failure to adhere to statutory requirements, such as establishing meaningful review procedures, could lead to regulatory actions by the Georgia Department of Insurance, including fines, cease-and-desist orders, or other enforcement measures available under existing insurance law. Similarly, public assistance agencies would likely face administrative consequences for failing to implement mandated human review processes. Furthermore, individuals negatively impacted by decisions made in sole reliance on AI, without proper human review, could potentially pursue legal avenues for redress, such as administrative appeals or civil litigation, arguing that the entity failed to comply with statutory requirements. The bill’s intent was to shift liability away from purely algorithmic decisions by mandating human accountability.

Relationship to Other Instruments

Georgia HB 887 was designed to amend existing sections of the Official Code of Georgia Annotated, thereby integrating its AI regulations directly into established legal frameworks. Specifically, it proposed amendments to Article 1 of Chapter 24 of Title 33, which pertains to general provisions regarding insurance; Article 1 of Chapter 34 of Title 43, related to the Georgia Composite Medical Board; and Article 1 of Chapter 4 of Title 49, concerning general provisions regarding public assistance. This approach indicates an intent to build upon and modify existing laws rather than create an entirely new, standalone regulatory regime for AI. By amending existing codes, the bill aimed to ensure that its provisions would be interpreted and enforced within the context of established legal principles and regulatory practices governing these sectors.

The bill's introduction also occurred in a broader context of AI regulation in Georgia. Notably, HB 887 followed the enactment of Georgia HB 203, which went into effect on July 1, 2023. HB 203 was a first-of-its-kind state law that permitted the use of AI tools in clinical settings, specifically for eye assessments and generating prescriptions, but crucially also required that AI data could not be the sole basis for issuing a prescription. HB 887, in contrast, sought to expand this principle of human oversight beyond optical care to general healthcare, insurance, and public assistance, indicating a legislative trend towards mandating human review for AI-assisted consequential decisions across various domains. This relationship suggests a developing, albeit piecemeal, state-level approach to AI regulation in Georgia, where subsequent bills build upon the principles established by earlier ones.

International Alignment

As a state-level legislative initiative within the United States, Georgia HB 887 primarily focused on domestic regulatory concerns and did not directly address international alignment. However, its core principles of mandating human oversight and prohibiting sole reliance on AI for consequential decisions align with broader global discussions and emerging regulatory trends in AI governance. Many international frameworks, such as the European Union's AI Act or recommendations from organizations like the OECD, emphasize the importance of human-in-the-loop mechanisms, accountability, transparency, and the prevention of algorithmic bias, especially in high-risk AI applications. The requirement for a "meaningful review" by an individual with override authority in HB 887 reflects these widely recognized principles of responsible AI development and deployment.

While not explicitly designed for international harmonization, the bill's approach contributes to a growing body of state-level AI regulations within the U.S. that collectively inform the national discourse on AI policy. These state efforts, including Georgia's, can influence federal legislative considerations and potentially contribute to a fragmented but evolving regulatory landscape that, in aggregate, may exhibit some thematic alignment with international best practices. The emphasis on sector-specific applications (healthcare, insurance) also mirrors a global trend of tailoring AI regulations to the unique risks and requirements of different industries. Thus, while not an international instrument itself, HB 887's principles resonate with global efforts to ensure ethical, safe, and human-centric AI systems.

Implementation Timeline

Compliance Checklist

Sources and References