Now that you've explored where AI can genuinely enhance your HR work, it's equally important to understand where it shouldn't be used. Knowing the boundaries of AI isn't about being overly cautious—it's about protecting your organization, your employees, and yourself from real risks. Throughout this unit, you'll learn the guiding principles that keep AI use responsible, recognize high-risk scenarios to avoid entirely, and identify safe starting points where you can confidently integrate AI into your workflow. Think of this as building your professional judgment muscle: understanding not just what AI can do, but when it's appropriate to rely on it.

Before using AI for any work-related purpose, you need to ground yourself in a few foundational principles. The most important rule is to avoid using AI for decisions or tasks involving the following:

• Confidential company data
• Personally identifiable information (PII)
• Legal interpretation
• Sensitive people-related issues

The only exception to this rule is if the use of AI tools for these applications have been explicitly approved under your company's AI policy. This means that before you copy and paste anything into an AI tool, you should pause and ask yourself whether the information would be appropriate to share externally.

Start by knowing and following your organization's AI usage guidelines. If your company hasn't formalized these yet, treat every piece of sensitive data as off-limits for unapproved tools. This includes employee social security numbers, salary details, customer information, and proprietary business strategies—none of which should ever be entered into AI systems that haven't been vetted by your IT or legal teams.

A helpful mental check is this: if you wouldn't email the information outside the company, don't put it into AI without explicit approval. Respecting confidentiality isn't just good practice; it's essential for maintaining trust and compliance with privacy regulations.

Here's an example of how this principle might play out in a real conversation between two HR colleagues:

• Jessica: I was thinking about using that new AI tool to help me summarize all the performance review comments before our calibration meeting tomorrow. It would save me hours.
• Dan: That sounds tempting, but have you checked if that tool is on our approved list?
• Jessica: I don't think so—it's just a free summarization tool I found online. Why does that matter?
• Dan: Performance reviews contain sensitive employee data. If you paste that into an unapproved tool, you could be exposing PII to a third party.
• Jessica: I hadn't thought of it that way. So what's the test I should use?
• Dan: Ask yourself: would I email this information outside the company? If not, don't put it into AI without explicit approval.

This exchange illustrates how easy it is to overlook data sensitivity when focused on efficiency. Jessica's instinct to save time is understandable, but Dan's simple question—"would I email this outside the company?"—provides a clear framework for making the right call in the moment.

Certain applications of AI in HR carry significant legal, ethical, and reputational risks. These are scenarios where human judgment, accountability, and oversight are non-negotiable.

If the previous section has you wondering where AI can be used safely, here's the reassuring news: there are plenty of low-risk, high-value applications to get you started.

In all these cases, the pattern is the same: use AI to accelerate your work, but keep a human in the loop for review and final decisions.

With these principles, risks, and safe starting points in mind, you're ready to practice applying your judgment. In the upcoming role-play session, you'll work through realistic scenarios where you'll decide when AI is the right choice—and when it's not.