State and local agencies are adopting AI for hiring, evaluations, and training. Most don't have the governance to manage it. Here's what to put in place now
Article Highlights
- Where state and local agencies are using AI in HR right now
- Why government AI use without a framework creates liability
- Building an AI governance framework for your agency
- How NEOGOV approaches AI governance in its own platform
Public sector HR is already using AI, but governance hasn't caught up. A 2026 NEOGOV survey of 4,200 public sector professionals found that 21% are actively using AI in their work, but only 28% of agencies have documented policies governing that use, and just 24% have trained employees on it.
These numbers reveal a gap in governance. Public sector HR has attempted to modernize without prioritizing AI oversight. Agencies use artificial intelligence for tasks like report writing, candidate screening, optimizing job descriptions, and performance documentation. But they haven’t created rules for when it's appropriate to use, what data can go into a prompt, or how to review outputs.
AI governance needs to be more stringent in the public sector, not less. Unlike the private sector, government HR needs to consider civil service rules, merit-based hiring requirements, EEO reporting, public records laws, and evolving state and federal standards.
State legislatures aren't waiting for agencies to catch up. Illinois, Colorado, and California have all passed AI employment laws that affect government hiring and personnel decisions, and federal guidance on AI keeps shifting as agencies try to maintain compliance. Your agency needs governance that can scale and evolve while mitigating liability today.
Closing the governance gap starts with awareness. Where is AI being used in your HR work? What liabilities is AI exposing you to? What does a defensible governance framework actually require? Keep reading to explore the answers to these questions.
If your agency hasn't implemented artificial intelligence yet, getting started with AI for government HR is a good place to begin before layering on formal governance.
Where State and Local Agencies Are Using AI in HR Right Now
AI adoption in government HR is concentrated in a few high-volume, high-burden areas where automation has the clearest efficiency payoff.
Job Posting Optimization
AI-assisted ad placement decides where your job postings run and how your recruiting budget splits across channels. AI can also analyze your job descriptions and recommend changes based on what attracts candidates in similar roles elsewhere. Some of these tools also flag language in a posting that could discourage qualified candidates from applying, which matters when your hiring process has to hold up against merit-based requirements.
Performance Evaluation Assistance
AI writing assistants help managers draft evaluations, summarize feedback, and structure comments into a complete review. Not only does that reduce the time managers spend on documentation, it makes the tone and thoroughness of evaluations more consistent across departments and supervisors.
Training Content and Compliance
AI-generated training content turns existing policy documents and training materials into quizzes, flashcards, and assessment questions. That removes the manual burden on training managers, who would otherwise build this content from scratch or update it every time a policy changes. PowerRecall, for example, generates flashcards directly from your agency's own policies and uses spaced repetition to reinforce retention, rather than relying on a one-time acknowledgment that an officer read a document.
In each case, AI should only assist the decision maker. It speeds up the work and your team still owns the outcome. That’s the core tenet of responsible AI in government HR, and one that NEOGOV’s AI solutions adhere to.
The Governance Gap: Why Government AI Use Without a Framework Creates Liability
Government agencies answer to civil service boards, auditors, elected officials, and the public. That scrutiny turns every ungoverned AI tool into a potential liability.
Merit-Based Hiring and Civil Service Compliance
Government hiring decisions must be defensible under civil service rules. If an AI tool influences candidate ranking or screening, your agency has to show that it doesn't violate merit-based hiring principles, veterans' preference, or EEO requirements. If you can't explain how a tool reached its recommendation, that recommendation is harder to defend in a grievance, audit, or lawsuit.
Public Records and Transparency Requirements
AI-assisted decisions can prompt public records requests and scrutiny from councils or boards. If your agency uses AI to draft documents or screen candidates, the methodology needs to be explainable and documented.
In Washington, two news outlets used public records requests to obtain city officials' ChatGPT logs, which showed staff using it to draft letters, constituent replies, talking points, and policy documents. North Carolina legal guidance treats prompts and outputs from government work as public records, even when a third-party platform hosts them, so staff should use agency-managed accounts.
State-Level AI Employment Laws
States are already writing AI into employment law, and doing it quickly. Illinois and California both define "employer" to include state and local governments, so public agencies fall within their rules.
- As of January 2026, Illinois holds employers responsible for discriminatory outcomes from their AI tools, and bars using zip codes as a substitute for protected traits such as race. Employers must also notify employees when they use AI in employment decisions.
- Colorado replaced its original AI Act in May 2026 with SB 26-189, effective January 1, 2027. It requires notice when automated tools are used in decisions about employment and other areas, and gives people the right to request human review of adverse outcomes.
- California's AI employment regulations, which took effect October 1, 2025, require employers to keep automated decision data for at least four years. They also warn that AI-driven assessments may count as unlawful medical inquiries if a test or question reveals a disability.
More states are likely to follow. Colorado's reversal before its first law took effect shows how fast requirements can shift, so your governance needs to anticipate where the law is heading rather than react to each new statute.
The Human Oversight Principle
In public sector HR, AI should never make decisions on its own, like rejecting a candidate or lowering a rating. Compliance guidance for public-sector HR identifies fully automated decisions like these as the biggest source of legal exposure.
Instead, make one person responsible for the results of each AI tool used in hiring, evaluations, or personnel decisions. That person checks the tool's output before anyone acts on it and has the authority to overrule it.
Building an AI Governance Framework for Your Agency
Public sector AI governance doesn't mean using AI less. It means deciding which tools staff can use, what they can use them for, and who checks the results. Here's a five-part framework for making that happen.
Define Approved Tools and Prohibited Uses
Start with an inventory of every AI tool your agency has approved, and note which ones meet your data security requirements. Then sort use cases into two lists: approved and prohibited.
The approved list could include drafting evaluation language, optimizing job postings, and generating training content. The prohibited list could include making final hiring decisions without human review or entering sensitive personnel data into an unapproved platform.
Publish both lists where staff can find them, and give people a way to request tools your agency doesn’t have. Otherwise, gaps in your tool set may push employees toward unapproved options.
Require Human Review for Every AI-Assisted Decision
AI-assisted decisions about hiring, evaluations, compensation, or discipline shouldn’t be finalized without a person's sign-off. Build that human review into the workflow. Your policy should state when the reviewer steps in and how the review is recorded, so you can prove that someone checked.
Build Bias Testing and Monitoring into Your Procurement Process
Before deploying an AI tool for hiring or evaluations, ask the vendor about its methods for bias testing, audit results, and the standards it measured against. Then write the contract to require ongoing bias audits and notification when the vendor updates its algorithm. Many agencies skip this step. In a 2026 survey of more than 600 state and local HR professionals, only 16% said their agencies test AI applications for bias.
Train Your Workforce on AI Use, Limitations, and Risks
Training has to reach everyone who touches an AI tool, not just leadership. Courses should cover what AI can and can't do, how to evaluate outputs for accuracy, what data never goes into a prompt, and when human judgment should override the tool. Repeat the training when tools or policies change.
Document Your Framework in Policy and Keep it Current
Write your AI governance framework as a policy, then distribute it to all relevant staff and collect electronic acknowledgments. Review it on a set schedule, as well as when state law changes or your agency adds a new AI use. Policy and procedure management software can handle distribution, version history, acknowledgments, and review reminders so the policy doesn't go stale.
How NEOGOV Approaches AI Governance in Its Own Platform
NEOGOV builds governance into its AI tools from the start, not as an add-on. AI is optional, and users have to actively opt in to use it. Once activated, AI only provides recommendations. Your HR team and managers review the output, edit if necessary, and make every final decision. Customer data isn't used to train public or global AI models, and it stays within your agency's environment.
Some safeguards don't involve AI at all. In Insight, PII blinding hides applicants' personally identifiable information on applications to support fair hiring. EEO reporting and diversity tracking show whether your applicant pool lines up with your hiring goals. These are standard platform features, not AI, but they give your team the visibility to catch problems early.
Your agency needs AI governance no matter which platform you use. NEOGOV builds the safeguards into the product so your HR team doesn't have to create workarounds to stay compliant. To see how, request a demo today or learn why government agencies choose NEOGOV.
