AI Hiring Compliance in 2026: NYC, Illinois, Maryland

As of April 14, 2026, the most practical U.S. AI hiring compliance work is not about chasing every AI headline. It is about knowing which rules actually apply to your hiring workflow, where they apply, and what your recruiting team needs to document before a tool is used on real candidates.

For most hiring teams, three jurisdiction-specific rules create the clearest operational obligations today: New York City's automated employment decision tool requirements, Illinois' video interview AI law, and Maryland's facial-recognition interview consent rule. On top of that, federal equal-employment law still matters when software screens, scores, ranks, or interviews applicants. If you run recruiting operations across multiple states, the safest approach is to build one repeatable review process and then add jurisdiction-specific controls where required.

This article is operational guidance for hiring teams, recruiters, and recruiting operations leaders. It is not legal advice.

Why this matters now

Hiring teams are using more software earlier in the funnel: resume ranking, chatbot screening, asynchronous video interview analysis, interview scoring assistance, and workflow automation that changes who advances. The risk is not only whether a vendor markets a feature as “AI.” The real question is whether the tool affects employment decisions, what notices or audits are required, and whether your team can explain how the tool is used in practice.

That matters because current U.S. compliance obligations come from a mix of city and state statutes, plus federal civil-rights law. A vendor saying a product is compliant is not enough on its own. New York City's FAQ states that employers and employment agencies are ultimately responsible for ensuring a bias audit was done before using an AEDT. EEOC guidance similarly makes clear that employers can face liability when automated tools create unlawful discrimination under existing federal law.

The three rules most hiring teams should know first

1. New York City: AEDT use can require a bias audit and advance notice

The New York City Department of Consumer and Worker Protection enforces Local Law 144 of 2021. Its FAQ says the law prohibits employers and employment agencies from using an automated employment decision tool in New York City unless they ensure a bias audit was done and provide required notices. The FAQ defines an AEDT as a computer-based tool that uses machine learning, statistical modeling, data analytics, or artificial intelligence, helps make employment decisions, and substantially assists or replaces discretionary decision-making.

For recruiting operations, the important trigger is broad: the FAQ says the law is not limited to final hiring decisions and includes screening. If you use a tool to meaningfully assess or screen candidates for a specific role tied to New York City, you need to evaluate whether the rule applies before using it.

The operational requirements are concrete:

A bias audit must be completed before use.
The employer or agency must publish a summary of the most recent bias audit and the tool's distribution date.
Notice to New York City residents must say that an AEDT will be used, identify the job qualifications or characteristics the tool will assess, include instructions for requesting reasonable accommodation, and be provided at least 10 business days before use.
A bias audit can only be relied on for one year from the audit date.

One detail many teams miss: the DCWP FAQ says the vendor is not responsible for your compliance. The employer or employment agency remains responsible for ensuring the audit happened before the tool is used.

2. Illinois: video interview AI triggers notice, explanation, consent, and deletion duties

Illinois' Artificial Intelligence Video Interview Act applies when an employer asks applicants to record video interviews and uses artificial intelligence analysis of the applicant-submitted videos for positions based in Illinois. The statute requires more than a generic privacy policy.

Before asking applicants to submit video interviews, an employer must:

Notify the applicant that AI may be used to analyze the interview and consider the applicant's fitness for the position.
Provide information explaining how the AI works and what general types of characteristics it uses to evaluate applicants.
Obtain consent before the interview.

The same Illinois act also limits video sharing. Applicant videos may be shared only with persons whose expertise or technology is necessary to evaluate fitness for the role. And if an applicant requests deletion, the employer must delete the interviews within 30 days and instruct anyone else who received copies to delete them as well, including electronically generated backups.

Illinois added another requirement in Public Act 102-0047. If an employer relies solely on AI analysis of a video interview to decide whether an applicant gets an in-person interview, it must collect and report specified race and ethnicity data annually to the Department of Commerce and Economic Opportunity.

3. Maryland: facial recognition in an interview requires applicant consent by waiver

Maryland's Labor and Employment section 3-717 is narrower than the New York City and Illinois rules, but it is easy to miss in practice. The statute says an employer may not use a facial recognition service to create a facial template during an applicant's interview unless the applicant consents.

That consent has to take the form of a signed waiver. The statute says the waiver must state in plain language:

the applicant's name,
the date of the interview,
that the applicant consents to the use of facial recognition during the interview, and
whether the applicant read the consent waiver.

If your interview stack includes facial analysis, identity, or emotion-recognition style features, this rule should be part of your vendor and workflow review for Maryland candidates.

Federal law still sits on top of these rules

Even where a city or state does not have a dedicated AI hiring statute, employers still operate under existing federal employment law. In 2022, the EEOC and DOJ warned that employers' use of AI tools can violate the Americans with Disabilities Act. EEOC guidance highlights several recurring risks, including tools that screen out people with disabilities, tests that measure traits in ways that disadvantage people with disabilities, and failures to provide reasonable accommodations when an automated tool is used.

The EEOC has also said Title VII applies to employers' use of automated systems to make or inform selection decisions and has pointed employers back to the Uniform Guidelines on Employee Selection Procedures when assessing adverse impact. In practical terms, that means a team should not assume that a vendor bias audit, a model card, or a four-fifths-rule summary is enough by itself. If your process uses software to help decide who advances, you still need a defensible, job-related hiring process and an accommodation path.

A simple compliance workflow for recruiting operations

The most reliable way to handle this is to review hiring technology at the workflow level, not feature by feature. Start with the actual candidate journey and mark every point where software materially changes what happens next.

Map the decision points. List where a tool screens, ranks, scores, recommends, or determines whether a candidate moves forward.
Match the role location. Check whether the position is tied to New York City, based in Illinois, or involves Maryland interview workflows.
Classify the tool's function. Separate resume parsing, interview scheduling, transcription, and note-taking from tools that assess, classify, or rank candidates.
Collect the compliance artifacts. Store bias audit links, notice language, consent language, accommodation instructions, and deletion workflows in one operating record.
Assign an owner. Someone in recruiting operations, people operations, legal, or security needs to own the review and annual refresh.
Re-check when the workflow changes. A tool that starts as transcription-only can become a scoring tool after a product update or a new setting is turned on.

If you manage hiring operations in a system like Apikuo, the useful pattern is to keep the compliance record attached to the workflow itself: which tools are active for a requisition, where notices live, when an audit expires, and how accommodation or deletion requests are handled.

Vendor questions worth asking before procurement or renewal

Does the tool make or materially influence screening, ranking, or advancement decisions, or is it strictly administrative?
For New York City use cases, do you provide an independent bias audit that an employer can rely on, and when does it expire?
What exact candidate characteristics or qualifications does the tool assess or infer?
Can we turn assessment features off by role, location, or workflow stage?
For Illinois video workflows, what applicant-facing explanation do you provide about how the AI works and what general characteristics it uses?
How are applicant videos shared, stored, and deleted, including backups?
Does any interview feature create a facial template or use facial recognition service technology?
What accommodation paths exist if a candidate cannot reasonably use the automated step?
What logs, reports, and retention controls are available to support an audit or investigation?

A 30-day checklist for hiring teams

Inventory every hiring tool that scores, ranks, classifies, or analyzes applicants.
Flag roles connected to New York City, positions based in Illinois, and interview workflows involving Maryland candidates.
Pull the current vendor documentation for bias audits, candidate notices, consent flows, and deletion controls.
Confirm whether any “AI assistant” or “analysis” setting is enabled by default in your interview stack.
Publish or review the notice and accommodation path required for New York City use cases.
Review Illinois candidate communications for notice, explanation, consent, limited sharing, and deletion handling.
Check whether any facial-recognition workflow exists for Maryland interviews and, if so, verify the waiver content.
Assign one owner to re-review these workflows whenever a vendor feature changes.

What to do next

If you only do one thing this quarter, stop treating AI hiring compliance as a general policy problem and treat it as a workflow-control problem. The laws with the clearest operational teeth today are tied to specific candidate experiences: screening in New York City, AI-analyzed video interviews in Illinois, and facial-recognition use in Maryland interviews. Once your team maps those points, the work becomes much more manageable.

The teams that handle this well usually do three things consistently: they know where automated assessment actually happens, they keep notices and audit records close to the hiring workflow, and they review vendor changes before those changes touch live candidates.

AI Hiring Compliance in 2026: What Recruiting Teams Need to Do in NYC, Illinois, and Maryland