HR is being asked to use AI. But what about I-9 workflows?
Your leadership team has asked you to incorporate AI into your HR workflows. So now, you use it for everything from answering benefits questions to recommending personal development paths.
But when it's time to onboard another cohort of hires, the stakes feel a little different. "Can AI really handle something as important as I-9 compliance?" you wonder, palms getting clammy.
You're right to be a little skeptical. I-9 compliance isn’t as simple as the form makes it look. Behind a few fields and signatures are federal laws, agency guidance, enforcement practices, and state requirements. Some tasks are an ideal use case for AI. Whereas others definitely require context, employee choice, or legal expertise.
So where should AI step in, and where should people stay in control? This guide draws the line and offers a practical framework for using AI in I-9 compliance.
This guide builds on a conversation between Immigration Attorney Kimberley Best-Robidoux and WorkBright CEO Chapelle Ryon. Want to hear more? Watch the full conversation here.
2026 changed the stakes for I-9 errors
If your team still treats some I-9 mistakes as “we’ll fix that if an audit ever finds them,” it’s time for a reset.
In March 2026, ICE updated its Form I-9 inspection guidance, reclassifying more than 10 categories of errors previously considered technical violations as substantive violations. Then, in August, ICE issued another update that put emphasis on electronic audit trails, work authorization changes, and penalty calculations.
The stakes are steep. Technical violations may generally be corrected after ICE identifies them during an inspection, but substantive violations don’t offer the same opportunity. Missing dates, document numbers, or other required information can now put employers at risk of penalties ranging from $288 to $2,861 per form, depending on the violation and penalty assessment.
The updates reinforce a critical point for employers using electronic I-9 systems: technology doesn’t transfer compliance responsibility. Employers are still accountable for meeting federal requirements, even when software handles the workflow.
As I-9 technology gets smarter, knowing what it’s doing, and whether it’s doing it correctly, matters more than ever.
AI is showing up on all sides of the I-9 process
Employers aren’t the only ones implementing AI. Immigration agencies are using technology to identify patterns, cross-reference records, flag potential risk, and prioritize enforcement activity. Cross-agency data sharing has also expanded, including an April 2025 agreement giving ICE access to IRS employer records.
ICE inspection activity has increased sharply as well. According to Greenspoon Marder’s analysis of immigration compliance statistics, the rate of Notices of Inspection in the first half of 2025 was at least 10 times higher than in 2024. The administration’s border czar has also proposed using contractors to expand inspection capacity.
At the same time, bad actors have access to AI. AI tools can make identity documents and credentials easier to manipulate. Deepfake technology can make it harder to determine whether the person appearing over video is actually the person presenting the document.
The 2026 documentary Infiltrated: North Korea’s Secret U.S. Workforce illustrates the scale of the problem. It followed a North Korean operation that applied to more than 1,000 U.S. companies in three months using stolen identities, AI-generated resumes, and face-swapping technology during live video calls.
The amount of digital document forgeries has increased year-over-year, according to the Entrust 2025 Identity Fraud Report.
I-9 rules don’t stop at the federal level
Keeping up with federal Form I-9 requirements is already a full-time job. Now, add state laws. Federal immigration rules establish the foundation, but employers increasingly have state-specific obligations layered on top. Those requirements can affect what employers must do during an ICE inspection, how they handle employee documents, and which employees must be run through E-Verify.
For example, Massachusetts’ PROTECT Act, signed August 5, 2026, requires employers to provide written notice to employees within 48 hours after receiving an ICE Notice of Inspection. California and Illinois have their own 72-hour notice requirements.
Colorado took a different approach. HB26-1283, effective June 3, 2026, created new requirements around employers temporarily holding employee passports and other identity documents. Employers must provide written notice of certain rights and retain the employee’s signed acknowledgment with the I-9. Violations can carry civil and criminal consequences.
E-Verify requirements are expanding in several states as well, with coverage varying based on factors such as employer size and industry.
And then there’s AI itself. States are also developing rules around automated decision-making, biometrics, data privacy, and disclosure when someone interacts with AI.
An AI system trained only to follow a federal workflow may miss a state-specific requirement. A general-purpose chatbot may confidently provide an answer without knowing which state’s rules apply.
That’s why an AI tool used in I-9 compliance needs more than a large, open knowledge base (like a standard LMM). It needs defined sources, defined boundaries, and a defined way to escalate questions when the answer isn’t clear.
Your I-9 vendor matters more than the AI label
“AI-powered” can sound impressive, but it doesn’t tell you much. Before asking what an I-9 vendor’s AI can do, start with the system underneath it. Does the software follow the rules for completing and retaining Form I-9? Does it preserve a defensible audit trail? Can your team access the information it needs when something goes wrong? Does the vendor keep its workflows current as requirements change?
A good I-9 platform should be able to show you how it handles requirements in the USCIS M-274 Handbook for Employers, 8 CFR 274a.2, applicable E-Verify requirements, electronic signature and retention rules, and the DOJ/DHS guidance on electronic I-9 systems.
It should also be able to explain what happens when something goes wrong. What happens when a case falls outside the normal workflow? Can you reconstruct who completed each step and when? Or what about when new requirements—like E-Verify v32—come out?
Those questions may not be as flashy as an AI demo, but they’re a lot more important.
Of decision-makers believe long-term business success depends on optimizing hiring with AI and automation, according to a study by Deloitte.
AI + humans: Who handles what?
I-9 compliance keeps going long after Section 2. Reverification deadlines come due, TPS designations change, and E-Verify cases stall while employees wait to act. Each area below shows what AI handles and what a human handles. When a case crosses from one to the other, AI should escalate it with the case details, its history, and a clear note on what action is needed.
Keeping cases moving
What AI handles
Every I-9 has deadlines. A new hire has to complete Section 1 by their first day, and Section 2 is due within three business days. E-Verify cases have deadlines too. AI can send reminders, watch for missed steps, and alert your team when a case stops moving.
What a human handles
Once AI alerts your team, a person takes over. This matters most when an employee gets a Tentative Nonconfirmation in E-Verify. Many employees think it means they’re about to lose their job. A person can explain what it actually means, help the employee decide whether to contest it, and check for simple mistakes, such as a typo.
Checking forms for mistakes
What AI handles
Many I-9 mistakes can’t be fixed once ICE starts an audit, so it’s better to find them yourself first. If you’re reviewing 200 I-9s, AI can check every form for missing or incorrect information and give your team a list of forms with problems.
What a human handles
AI finds the mistakes, but a person fixes them. Someone on your team looks at each form AI flagged, decides whether the mistake can be corrected, and makes the correction. The correction has to be clearly marked on the form. AI should never change a form on its own.
Checking documents
What AI handles
Fake documents are getting harder to spot. AI can look for signs that a document was changed, compare the photo on an ID with a selfie, and confirm the person is really there. This gives your team a first check before anyone reviews the document.
What a human handles
A person still has to review the documents for every new hire, not just the ones AI flags. The person who completes Section 2 signs it under their own name. If AI flags a document, or your team sees a document it doesn’t recognize, a person looks into it. They may need to ask an immigration attorney or the Department of Justice’s Immigrant and Employee Rights Section for help. AI should never reject a document on its own, because that could lead to discrimination.
Tracking expiration dates and reverification
What AI handles: Work authorization dates can change after you complete the I-9. USCIS ended the 540-day automatic EAD extension for I-765 applications filed on or after October 30, 2025. H.R. 1 capped TPS-based EADs at one year or the length of the TPS designation, whichever is shorter, which cut some extensions off at July 22, 2026. Plenty of I-9s still showed the later date.
Finding every affected form by hand is slow, and it's easy to miss one. AI can track expiration dates across your workforce and find them all at once. Some of the information you need doesn’t appear on Form I-9, so AI can also read the category and country of origin from each EAD card and show you which employees fall under which TPS designation. It can then handle outreach to the affected employees.
What a human handles: The reverification itself stays with a person. An employee may qualify for an extension only with certain paperwork, or may bring several documents without knowing which one applies. The employer can’t choose a document for the employee, so a person has to explain the options and let the employee decide. A person also completes Supplement B and every other legally required step.
Prioritizing the work
What AI handles: With thousands of I-9s in a system, your team can’t watch every record. AI can sort cases by urgency, separating the ones that can move through a standard workflow from the ones that need a person. It can also pull a case’s history into one view, so a reviewer doesn’t have to open five screens to reconstruct what happened.
What a human handles: A person evaluates each exception the AI surfaces and decides the next step.
Supporting employees
What AI handles: Employees often get stuck outside business hours, and HR isn’t available around the clock. AI can answer common questions at any hour, walk employees through completing the I-9, guide them through submitting documents, and explain what happens next. It should be trained only on vetted I-9 guidance, such as the M-274 Handbook for Employers, and cite its sources so employees can check the answer. It must never misstate an employee’s document options or suggest anything about their eligibility.
What a human handles: A question should go to a person once it goes beyond the standard steps, or the moment the AI is unsure. An employee with a document problem needs someone to help resolve it. An employee with an unusual situation, such as a recent legal name change, needs someone who can ask the right follow-up questions. An employee who doesn’t understand why a rule applies to them needs someone to explain the requirement. And when the workflow can’t move a case forward, a person has to troubleshoot it with the employee.
Making employment decisions
What AI handles: AI supplies information, such as an alert that a work authorization date has passed or that an E-Verify case has closed, but it makes no decisions here.
What a human handles: Eligibility determinations, decisions on Tentative and Final Nonconfirmations, and any action to delay a start date, suspend an employee, or end employment all belong to the employer. A person has to confirm the facts and decide what happens next. That review matters most in E-Verify: an employer can’t take action against an employee because of a Tentative Nonconfirmation while the employee is contesting it.
What to ask your I-9 vendor
Before you get into model architecture, training data, or AI agents, make sure the I-9 system itself is built correctly.
- Can you show me how you meet the requirements?
Ask the vendor to demonstrate how its workflows align with the USCIS M-274 Handbook for Employers, 8 CFR 274a.2, the DOJ/DHS guidance for electronic I-9 software, and applicable E-Verify requirements.
- Can we access our E-Verify account and status change reports?
E-Verify’s status change reports add another layer of information employers need to monitor. Ask how the platform surfaces those reports and how it handles current E-Verify requirements.
- How do you document and audit every step?
You should be able to reconstruct who did what, and when, for any individual form. Ask to see the audit trail rather than taking the vendor’s word for it. Then get specific about AI.
- What tasks does the AI actually perform?
“AI-powered” is too broad. Does it launch workflows? Review forms? Answer questions? Flag documents? Track deadlines? Get specific.
- What is it trained on?
Ask whether the AI relies on official USCIS and government guidance, vetted internal processes, or open-web content.
- When does the AI escalate to a person?
Define escalation triggers, not just a generic “contact support” option.
- How is I-9 data protected?
Understand where employee identity and work authorization data goes when AI processes it, who can access it, and what is retained.
- What happens when the AI is wrong?
Ask how the vendor identifies errors, documents them, corrects them, and prevents the same issue from affecting other forms.
- Who owns the outcome?
The answer should be clear before you deploy the technology.
What AI + humans looks like in practice
Okay, you got us. We know all of this for a reason. We’ve spent years developing compliant AI-powered I-9 technology for WorkBright+, our full-service I-9 management offering.
WorkBright+ pairs AI agents with trained HR compliance specialists to manage the I-9 process from start to finish. AI handles the volume and consistency of routine tasks, while specialists step in whenever a situation calls for human expertise. Your team retains control over every employment decision.
Built in collaboration with legal counsel, WorkBright+ establishes clear boundaries between what AI can handle and where a specialist needs to step in. Together, they support everything from Section 2 completion, document review, and fraud detection to E-Verify case resolution, reverification, and employee support. AI also prioritizes daily tasks, tracks work authorization deadlines, and helps employees navigate routine questions using trusted compliance resources. When a situation requires judgment or falls outside the AI's approved scope, a specialist takes over.
The result is a full-service I-9 solution that combines the efficiency of AI with the judgment and support of experienced HR compliance specialists. AI does the work it's designed to do, people handle the situations that need a human touch, and your team stays in control.
Want to see what that looks like for your team? Explore WorkBright+ to learn how AI-powered technology and dedicated compliance specialists can take I-9 management off your plate.
