HB672 is a broad information-protection and artificial intelligence governance bill that would significantly expand the Commonwealth Office of Technology’s authority over how executive branch agencies use AI. It defines a range of AI-related terms, including generative AI, high-risk AI, foundation models, and synthetic media, and directs the Office of Technology to establish policy standards for responsible, ethical, and transparent AI use. The bill creates an Artificial Intelligence Governance Committee, requires a centralized registry of AI systems used by state government, and sets up an approval process for internal and external AI use cases. It also requires agencies to document intended uses, assess risks, ensure human review, and provide public disclosures when AI is used in decisions affecting citizens or businesses.
The bill also imposes detailed operational requirements on state agencies and third-party vendors. Agencies would need to show that AI is the best solution before using it, maintain impact assessments for high-risk systems, limit data use to what is necessary, protect personal data, and avoid using unexplainable AI in decision-making or approval/denial processes. Third-party AI developers, providers, and contractors would be subject to review by the Office of Technology, including scrutiny of testing, security, bias risks, documentation, and compliance with recognized standards such as ISO/IEC 42001. The bill further requires annual reporting to legislative committees on AI inventories, approvals, and vendor reviews, and it directs the Office of Technology to issue regulations to implement the new framework.
In addition to AI governance, HB672 amends Kentucky election law to address synthetic media in electioneering communications. It defines synthetic media and gives election officers or candidates a private right of action to seek injunctions, damages, attorney’s fees, and costs against sponsors of electioneering communications that use altered images, audio, or video without a clear disclosure. The bill places the burden of proof on the plaintiff, limits liability for the medium disseminating the communication in most cases, and preserves certain protections for interactive computer service providers while allowing liability in narrow circumstances. These election-related provisions are aimed at preventing deceptive deepfakes and other manipulated media in campaigns.
The bill’s impact on state law would be substantial, especially for executive branch technology procurement, data governance, and administrative decision-making. It would centralize AI oversight in the Commonwealth Office of Technology, create new compliance duties for agencies, and establish new transparency and reporting obligations. It would also add a new cause of action under election law for synthetic media used in political communications, potentially affecting campaigns, advertisers, media outlets, and online platforms. Because the bill declares an emergency, it would take effect immediately upon enactment.
No committee transcripts or recorded votes were provided, so there is no direct evidence of debate or formal support/opposition in the available materials. Based on the bill text alone, the overall tone appears strongly pro-regulation and precautionary: it emphasizes privacy, security, transparency, human oversight, and anti-discrimination safeguards. Likely points of contention would include the breadth of centralized state control over AI use, the administrative burden on agencies and vendors, the feasibility of compliance with detailed documentation and approval requirements, and the scope of liability for synthetic media in election communications.
HB672 would amend Kentucky law governing executive branch information technology and create a new statutory framework for artificial intelligence oversight, including centralized approval, inventory, reporting, vendor review, and public disclosure requirements. It would also add election-law remedies targeting synthetic media in electioneering communications, creating potential civil liability for sponsors and limited liability rules for media and online intermediaries. The bill would primarily affect state agencies, the Commonwealth Office of Technology, AI vendors and contractors, candidates, election officers, advertisers, and media/platform operators.
No votes or committee discussion were provided, so the recorded sentiment cannot be measured directly. The bill’s text reflects a generally cautious and protective posture toward AI and election integrity, with a strong emphasis on privacy, transparency, human oversight, and fraud prevention. That suggests the bill is framed as a public-safety and governance measure rather than a deregulatory or innovation-promoting one.
The main likely points of contention are the bill’s broad centralized control over AI use in state government, the detailed approval and reporting requirements, and the prohibition on unexplainable AI in consequential decisions. Agencies and vendors may view the compliance burden as significant, especially given the required impact assessments, registry entries, and third-party reviews. The election provisions may also be contested over free-speech concerns, the scope of liability for sponsors and media outlets, and how clearly synthetic media and disclosure requirements can be applied in practice.