H.752 would expand the Agency of Digital Services’ oversight of automated decision systems used by Vermont state government. The bill requires the Agency to conduct and maintain a detailed annual inventory of all automated decision systems being developed, procured, or used by state government, including whether the systems are final-decision tools or support tools, what data they use, whether they have been tested for bias, how they are secured, and what privacy, safety, and fiscal impacts they may have. It also requires the annual IT and cybersecurity report to include updates on the inventory and any instances where the Agency asked a state entity to stop using a system.
The bill also gives the Agency authority to request termination of a state government entity’s use of an automated decision system if the inventory review finds that the system produces biased results, poses unreasonable safety or cybersecurity risks, is likely to be misused without adequate safeguards, or does not properly protect Vermonters’ personal data. The bill defines key terms such as algorithm, automated decision system, automated final decision system, and automated support decision system, and it would take effect on July 1, 2026.
Its impact on state law would be to add a more formal review and reporting framework for government use of artificial intelligence and other algorithmic tools, particularly in procurement and deployment decisions. It would increase state oversight of vendors and agencies using automated systems, and it could lead to the discontinuation of certain technologies if they are found to be biased, insecure, unsafe, or privacy-invasive. The bill would also create a clearer statutory basis for tracking the costs, benefits, and risks of these systems across state government.
Because there are no committee transcripts or recorded votes provided, the overall sentiment cannot be measured from legislative debate or roll call history. Based on the bill text alone, the measure appears to reflect a precautionary and oversight-oriented approach to AI in government, emphasizing cybersecurity, privacy, and fairness. The absence of recorded opposition or support in the provided materials means no specific political alignment can be inferred beyond the bill’s protective intent.
The main points of contention likely would center on how broad the Agency’s review authority should be, how burdensome the inventory and reporting requirements may be for agencies and vendors, and whether the termination-request authority is too strong or too vague. Potential supporters would likely emphasize transparency, consumer protection, and risk reduction, while critics might raise concerns about administrative burden, procurement delays, and whether the standards for bias, misuse, or cybersecurity risk are sufficiently objective.
The bill would amend Vermont law governing the Agency of Digital Services by expanding annual reporting requirements and by requiring a comprehensive inventory of state government automated decision systems. It would also authorize the Agency to request that a state entity terminate use of a system identified as biased, unsafe, insecure, misused, or insufficiently protective of personal data, thereby increasing state oversight of AI and algorithmic tools used in government operations.
No committee discussion or vote history was provided, so there is no recorded legislative sentiment to summarize from debate or roll calls. From the text, the bill appears generally protective and cautionary, reflecting concern about bias, cybersecurity, privacy, and misuse in state government use of automated systems.
Likely areas of contention include the scope of the Agency of Digital Services’ authority, the administrative burden of detailed inventories and annual reporting, and the practical effect of the Agency’s power to request termination of a system’s use. Supporters would likely favor stronger oversight and safeguards for Vermonters’ data, while opponents may worry about procurement delays, compliance costs, and the subjectivity of determining bias, misuse, or unreasonable risk.