What SB 2909 actually bans
Governor JB Pritzker signed Senate Bill 2909 on July 10, prohibiting school administrators from using AI to assign a numerical score or qualitative rating to any component of a teacher's performance evaluation. The law, effective January 1, 2027, closes off the most obvious shortcut an AI vendor could pitch to a resource-strapped district: let the software do the scoring so an overworked principal does not have to sit through as many classroom observations. State Senator Christopher Belt, the bill's sponsor, was direct about the rationale: "I believe that our teachers should be judged based on actual observations and professional judgment, not by AI software."
The law also restricts teachers themselves: they cannot use AI to produce the written self-assessments or evidence components that go into their own evaluation file. Belt framed the dual restriction as a fairness measure: "Our educators deserve a transparent and fair evaluation process that demonstrates their actual work in the classroom and protects their privacy." The law does carve out room for AI in purely administrative evaluation tasks, scheduling observations, compiling non-judgmental records, as long as both the teacher and evaluator disclose which tool they used and why.
The unanimous vote that should get a vendor's attention
SB 2909 passed both chambers of the Illinois General Assembly without a single recorded no vote. Unanimity on any AI legislation is rare enough to be the actual headline here, more than the specific restriction itself. Lawmakers who disagree on nearly every other issue found common ground on keeping AI out of a process that determines a teacher's job security, which suggests this was a restriction legislators saw as obviously necessary once the alternative was spelled out, rather than a partisan fight anyone wanted to pick.
For any vendor who assumed AI-assisted evaluation tools would face the normal mix of enthusiastic early adopters and cautious skeptics typical of ed-tech procurement, a unanimous legislative ban is a different category of signal. It means the political cost of defending an AI-scored evaluation product in Illinois, or explaining to a school board why your tool should be exempted, is now effectively zero upside and real downside. That calculus will shape how aggressively similar vendors market evaluation features in every other state watching Illinois's session.
The guidance that came the day before
One day before Pritzker signed SB 2909, the Illinois State Board of Education released statewide guidance built around four principles: human relationships shape learning, schools serve academic, developmental, and civic purposes beyond test scores, AI should inform rather than replace teaching, and implementation decisions should be made locally rather than mandated from the state down. The guidance explicitly states that "AI should support, rather than replace, educator expertise and professional judgment," language that reads as a direct companion to the evaluation ban signed the next day.
The guidance covers more ground than the law does: instructional use, district readiness, governance structures, AI literacy for students and staff, equity, and accessibility. It encourages teachers to use AI for lesson planning and assessment development, the kind of preparation work several other states have already blessed, while drawing a clear line at anything involving a direct judgment call about a specific student or employee. Illinois is effectively building a two-track policy: AI welcome for preparation work, AI excluded from judgment calls about people.
Why evaluations specifically, and not AI in the classroom generally
Illinois did not ban AI from classrooms, curriculum design, or administrative workflows. It drew the line specifically at evaluations, the one process in a school system that most directly determines pay, tenure, and continued employment. That specificity is the lesson for any technology leader watching how AI regulation actually gets written: lawmakers are not reaching for broad AI bans, they are identifying the single highest-stakes decision point in a given domain and restricting AI there first, while leaving lower-stakes AI use largely alone.
That pattern should look familiar to anyone who has watched AI regulation emerge in hiring, lending, or healthcare over the past three years. Credit decisions, hiring decisions, and now teacher evaluations are converging on the same regulatory logic: AI can assist the process, but the final judgment about a person's livelihood has to trace back to an accountable human, documented in a way that survives a legal challenge. Education just became the latest domain to formalize that line into statute.
What this means for HR-tech and performance-management vendors
Any company selling AI-assisted performance evaluation, scoring rubrics, or HR analytics into K-12 schools needs to rebuild its Illinois go-to-market around what is left: administrative efficiency, scheduling, non-judgmental documentation, with mandatory disclosure of which tool was used and why. Features that score, rate, or generate evaluation narratives are no longer viable in that state regardless of how they are marketed, and generic "AI-assisted insights" framing will not survive a legal reading of SB 2909's plain text.
The more important question for a vendor is what to change everywhere, not just in Illinois, because a unanimous vote in one state legislature is the kind of precedent other states' education committees cite directly when drafting their own bills next session. Vendors with evaluation-scoring features anywhere in their product line should treat Illinois as the floor for what survives contact with a legislature, and plan for the ceiling of regulatory scrutiny to keep rising from here.
The pattern forming across states
SB 2909 was one of several AI-related bills Illinois passed this year, alongside the Artificial Intelligence Safety Measures Act, which requires transparency obligations from companies including OpenAI and Anthropic. Taken together, Illinois is building a broader AI governance framework piece by piece rather than through one sweeping statute, and education is just the sector that happened to get the clearest, most specific rule first. Expect labor, healthcare, and consumer lending to be next in line for the same treatment, since the legislature has now shown it is comfortable regulating AI domain by domain rather than waiting for a single comprehensive bill to cover every use case at once.
For a CTO or CIO tracking regulatory risk across multiple states, the practical takeaway is to stop waiting for a comprehensive federal AI law and start treating state-by-state, domain-specific bans like this one as the actual regulatory environment you are operating in now. Illinois will not be the last state to decide that AI has no business making the call on whether someone keeps their job, and the next states to move will likely borrow this bill's language almost directly rather than starting from scratch. Build your vendor contracts and internal AI policies assuming more states follow this pattern, and assuming the specific carve-outs Illinois wrote in, disclosure requirements and administrative-task exceptions, become the template other legislatures reach for first.
