HB3567 creates the “Meaningful Human Review of Artificial Intelligence Act” and would regulate how Illinois state agencies use automated decision-making systems, including artificial intelligence and algorithmic tools. The bill generally bars a state agency, or anyone acting on its behalf, from using such systems without continuous meaningful human review when the system is used for public assistance benefits, for actions that materially affect rights, civil liberties, safety, or welfare, or for decisions affecting statutory or constitutional rights. It also limits state procurement of systems that would be used for those purposes unless they are subject to that same human review.
The bill defines “meaningful human review” as oversight by trained individuals who understand the system and can approve, deny, or modify its output. It also excludes ordinary office software and internal administrative tools that do not materially affect individual rights or benefits. In addition, the bill protects existing employee and collective bargaining rights by stating that use of automated systems cannot displace workers, transfer their duties to automation, or alter civil service or bargaining status.
Before using a covered system, agencies would have to complete an impact assessment and repeat it at least every two years and before any material change to the system. The assessment must describe the system’s objectives, explain how it was developed, summarize algorithms and training data, and test for accuracy, fairness, bias, discrimination, cybersecurity, privacy, public health and safety risks, misuse, and handling of sensitive personal data. If an assessment finds discriminatory or biased outcomes, the agency must stop using the system and any information produced by it.
The bill would also require agencies to submit impact assessments to the Governor, the President of the Senate, and the Speaker of the House at least 30 days before implementation, and to publish the assessments online. If enacted, it would create new compliance duties for state agencies and likely slow or condition deployment of AI-driven tools in public programs and other high-impact government functions.
Because there are no committee transcripts or recorded votes provided, the public sentiment cannot be measured from debate history. Based on the bill text alone, the measure appears strongly precautionary and oversight-oriented, reflecting concern about bias, discrimination, privacy, and accountability in government use of AI. The main points of contention likely would be whether the human-review requirement is too restrictive for agencies, how broadly the covered functions are defined, and whether the bill could limit efficiency or innovation in state operations.
HB3567 would add a new state-law framework governing automated decision-making by Illinois agencies. It would impose a prohibition on using covered AI or algorithmic systems without continuous meaningful human review in high-impact contexts, require recurring impact assessments, mandate public reporting and legislative submission, and require agencies to stop using systems found to produce discriminatory or biased outcomes. It also preserves existing labor and collective bargaining protections by preventing automation from replacing state employees or changing their employment rights.
No committee discussion or vote history was provided, so there is no recorded legislative sentiment to summarize from debate or roll call. The bill’s text suggests a generally cautious, pro-regulation approach to AI in government, emphasizing human oversight, transparency, and anti-discrimination safeguards. The overall posture is protective of residents and workers rather than permissive toward automation.
The likely points of contention are the scope of the ban and the operational burden on agencies. Supporters would likely favor the bill’s safeguards against biased or opaque AI decisions, especially in public benefits and rights-affecting decisions, while critics may argue that the continuous human-review requirement, repeated impact assessments, and mandatory shutdown of systems with biased outcomes could be costly, slow implementation, and limit beneficial automation. Labor-related provisions may also draw attention because they explicitly restrict using automation to displace employees or transfer duties.