Here’s a scenario I’ve run into more than once. We onboard a new client, ask to see their employee handbook, and they say yes, we have one. When we ask how it was put together, the answer is often the same: “I pulled it from ChatGPT.”
I’m not an employment attorney, and neither is ChatGPT. That handbook probably doesn’t reflect your state’s leave laws. It almost certainly doesn’t account for how your organization actually operates. And depending on when it was generated, it may reference guidance that has since been superseded. But it looks polished, it’s formatted nicely, and it filled a gap quickly. I understand the appeal.
That’s where a lot of organizations are right now with AI in HR. The tools are genuinely useful, and they’re also being treated as more authoritative than they are. For HR, where the output carries real legal and compliance weight, that gap matters.
Where AI Genuinely Helps HR Teams
I use AI regularly, and it has changed how I work. What used to take two hours to compile and format, I can now have in draft form in minutes. For lean HR teams, there are real, low-risk places where AI adds value.
Drafting job descriptions and policies. AI is excellent at getting you from zero to a starting point. A blank page disappears quickly, and you have structure and language to react to. That’s genuinely helpful when you’re managing a full plate.
Learning and development content. Building onboarding materials, training outlines, and internal communications moves faster with AI assistance. For small organizations without a dedicated L&D function, this can make programs possible that otherwise wouldn’t exist.
HR platform chatbots. Many payroll and HR platforms now use AI to handle routine employee questions around PTO balances, benefits enrollment, and pay timelines. When these are well configured, they reduce administrative volume and give employees faster answers.
Data analysis and reporting. This one is underutilized. If you pull turnover data from your HR system and run it through a secure AI platform, you can get analysis and visualization in a fraction of the time it would take manually. It surfaces patterns that might otherwise stay buried in a spreadsheet.
These are real productivity gains, and I don’t want to dismiss them. The issue isn’t that organizations are using AI. It’s what happens, or doesn’t happen, after.
The Real Risks of Using AI in HR Without Oversight
The most common mistake I see is treating AI output as final. Whatever the tool produces gets copied, approved, and put into use without meaningful review. That approach works fine for low-stakes content. For HR, it almost never does.
A few specific risks worth understanding:
AI doesn’t know your state. This is the one I have come back to most. If you use an AI tool to draft a leave policy, a termination procedure, or an employment classification framework, that tool is not referencing the applicable laws for your state or the city where your employees work. It is generating plausible language based on patterns. Maryland’s family leave requirements are different from California’s, which are different from Massachusetts’. A policy that looks complete may be missing critical compliance elements specific to where you operate.
The handbook problem is more common than you’d think. I’ve seen organizations go years with a handbook that was never reviewed by anyone with legal knowledge, HR expertise, or even a basic familiarity with their own policies. The document exists, which gives leadership a false sense of security, but it hasn’t been validated against anything that actually governs how the organization should operate.
Data privacy deserves more attention than it gets. HR manages some of the most sensitive employee information that exists: Social Security numbers, compensation data, medical information, immigration documentation. There is a significant difference between a company-approved, IT-secured AI environment and an employee using whatever’s available on their desktop. If your team is using public or unapproved tools for HR tasks, you may be exposing confidential information without realizing it.
Recruiting and screening AI carries bias risk. We’re starting to see early signs of AI being used for candidate screening among some of the clients I work with. As this use increases, it needs to be approached carefully. If AI is filtering or ranking candidates, someone must understand the criteria it’s using, audit it regularly for patterns or bias, and be able to clearly explain those decisions. This is also an area drawing growing regulatory attention, with some states already requiring transparency around the use of AI in hiring.
What Responsible AI Use Actually Looks Like
None of this means you shouldn’t use AI. It means you need a clear-eyed process for what comes next.
Prompt for sources, then verify them. When I use AI for anything involving regulations or compliance guidance, I explicitly ask it to provide source links. Then I click through. If it takes me to the Department of Labor or a state agency, I know I’m working from something legitimate. If it can’t cite sources, or the links don’t hold up, that’s a signal to dig further before relying on the output.
Build a review step into your workflow before anything becomes official. For job descriptions, policies, handbook language, or any documentation employees will rely on, create a checkpoint where a qualified person reviews before it’s finalized. This doesn’t need to be elaborate. It just needs to exist.
Keep legal review in the loop for anything that becomes policy. I always tell clients: anything going into your handbook or becoming official company policy should have a legal review before it’s your source of truth. AI can help you draft it, but an employment attorney should be the last set of eyes.
Know what your people are actually using. One of the most important questions HR leaders can ask right now is what AI tools employees are using for work, and where that data is going. At SC&H, we have approved platforms and protocols in place. Most of our clients don’t have anything established yet, which means employees are making individual decisions about what to use and potentially putting sensitive information into tools the organization has no visibility into.
The Right Way to Think About It
AI is good at getting you from zero to something. It is not the final answer. The organizations that use it well treat it like a capable first draft: useful, fast, and always in need of someone with context and accountability to take it the rest of the way.
For HR in particular, that human element matters more than in almost any other function. The decisions HR touches — who gets hired, how employees are classified, what policies govern people’s work, leave, and pay — have real consequences when they’re wrong. AI doesn’t carry that accountability. Your organization does.
If you’re not sure how AI is currently being used across your HR processes, or whether the content it’s generated has ever been properly reviewed, that’s often the first signal it’s worth taking a closer look. Our team works with organizations to identify where these gaps exist and put the right practices in place before they become a problem.
