Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Ai Regulation topic
No spam. Unsubscribe anytime.
Montana committee hears bill to limit government use of AI; DOJ urges narrow law‑enforcement carveouts
Summary
House Bill 178 would restrict state and local government uses of artificial intelligence—banning certain surveillance, behavioral‑manipulation and discriminatory classification, requiring disclosure of AI‑generated materials and mandatory human review; the Department of Justice voiced 'soft opposition' seeking targeted exemptions for investigative uses.
Get email alerts on the Ai Regulation topic
No spam. Unsubscribe anytime.
Helena — The House Judiciary Committee on [date] heard testimony on House Bill 178, a proposal from Representative Braxton Mitchell (House District 5) to regulate state and local government use of artificial intelligence. Mitchell told the committee the bill aims to "limit misuse, ensure transparency, and preserve human oversight" by prohibiting certain applications, requiring disclosure when governments publish AI‑generated content and mandating human review of automated recommendations that could affect a person's rights.
Mitchell, the bill sponsor, said the measure would bar uses such as behavioral manipulation, unlawful discrimination, malicious actions and certain types of surveillance. He told members he planned to circulate an amendment that defines "behavioral manipulation" and said he is open to working with the Department of Justice on a law‑enforcement exemption.
The Department of Justice, represented by Alex Sturhan, director of external affairs, said DOJ was in "soft opposition" to the bill as drafted. Sturhan told the committee that some law‑enforcement applications—such as scalable tools to support human‑trafficking investigations, profile generation and investigative data analysis—could be useful in Montana's rural context and that a targeted exemption for those investigative uses likely would change DOJ's position.
Informational witnesses from state technology offices described current and planned uses of AI. Kevin Gilbertson, Montana's state chief information officer, said the state is piloting AI to analyze roadway camera images and automate updates to the state's 511/501 traffic sites. Adam Carpenter, chief data officer for the state, distinguished routine analytic or machine‑learning tools (for example, traffic‑pattern detection or recommendation engines) from autonomous systems that make independent decisions, and said the bill's intent should be to prevent discriminatory or manipulative uses rather than to ban benign analysis.
Committee members pressed the sponsor and witnesses on several practical risks and scope questions. Lawmakers asked whether the bill's prohibition on "classifying a person or group based on behavior, socioeconomic status or characteristics resulting in disparate impact" could unintentionally block legitimate forecasting and planning work, such as fiscal office projections or HHS forecasting of Medicaid needs. Witnesses and the sponsor said intent and language matter: defensive phrases in the bill intended to target malicious or deceptive uses should leave room for neutral, aggregate analytics used for planning and public‑safety monitoring.
Members also raised concerns about automated surveillance and facial‑recognition‑style profiling. Carpenter and Gilbertson said the sponsor appears focused on preventing systems that build persistent corpora of faces for tracking individuals and other invasive functions, while allowing traffic‑pattern detection and limited, lawful uses to locate missing persons or assist investigations under Title 44 exceptions.
The hearing featured extended discussion about bias and unintended consequences. Witnesses confirmed that AI systems can reproduce and amplify biases in training data and gave examples (such as medical imaging or prior industry experiences) where models missed or misclassified outcomes because of skewed or incomplete datasets. They emphasized practitioner responsibilities: data preparation, normalization and human review are central to mitigating those risks.
Several lawmakers asked whether the legislature should study AI further before passing law. Mitchell argued for statutory guardrails now, noting a similar bipartisan bill had passed the New Hampshire House; others urged careful drafting to avoid sweeping language that would capture ordinary state analytics, website AB‑testing or public‑health advisories.
No formal committee action or vote occurred at the hearing. Mitchell closed by reiterating his willingness to work with DOJ and other stakeholders on tailored amendments, and the committee chair closed the hearing and reminded members that executive action on bills is scheduled for the next day.
What comes next: The committee will consider amendments during executive action. Committee members signaled support for clarifying language about malicious intent and for working out a law‑enforcement exemption that preserves investigative capabilities while guarding against discriminatory or surveillance abuses.
