An HR analyst has built a side practice in AI prompt engineering, billing $400 an hour on evenings and weekends. She never told her employer, which has no moonlighting policy. Her manager found the practice through a LinkedIn post and raised three concerns: loyalty, distraction, and whether she used a company-licensed AI tool for personal gain. On HR Voices, Miriam Chan, Head of Global People Operations at Miro, told host Rebecca Taylor where she would begin.
The brief asks what the company can restrict. Rebecca questioned whether restriction was the goal at all. Miriam started with the manager, not the analyst, and treated a new policy as the last step. Her order is discovery, then the problem, then the rule.
Why the Manager’s Motivation Comes First
What Miriam found least clear was “the manager’s motivation in bringing this up.” Something had led the manager to look through a direct report’s LinkedIn. A performance issue could sit underneath. Privacy rules also depend on where both people are located.
The stated concerns were thin. Miriam saw nothing in “loyalty” that anyone could prove. As long as the analyst performs well, she said, “work is also work.” Rebecca noted that this is white-collar work, not an hourly job, so there was no clock to be on. Without a performance problem, distraction is a prediction, not a finding. The analyst never disclosed the practice, but she posted it openly. Miriam suspected “more to the story that the manager is not sharing.”
What HR Can Restrict When There Is No Policy
Side work is common. The Bureau of Labor Statistics counted 8.8 million multiple jobholders in August 2026, 5.4 percent of everyone employed. This company never wrote rules on moonlighting, and Miriam read the gap as information. A missing policy, she said, “also says something about the organization.”
The firmer ground is acceptable use. Many of those policies “were written before we had AI tools,” Miriam said. Gallup found in 2025 that only 30 percent of U.S. employees said their organization had general guidelines or formal policies for using AI at work. If the analyst ran client work through a company license, Rebecca saw a company property problem with a clear fix. Both called it fair to limit AI use on unrelated work, and Miriam has heard of companies capping it because tokens are expensive.
The checks raise questions of their own. Reading the analyst’s AI history is a privacy matter, so Miriam would look only where policy allows it. Rebecca would fold a social media use policy into the review, since a LinkedIn post is still social media.
Whatever HR decides “sets a precedent,” Miriam warned, because no rule existed before. Punishing the analyst could also backfire. “We actually want our employees to be using AI to be successful,” Miriam said, and policing one person’s side work can discourage everyone else from exploring AI.
What Actually Works
Start With Discovery, Then Define the Problem
Miriam calls hers a “product manager approach to HR”: discovery first, then a definition of the actual problem. She asks the manager what prompted the concern, what the word loyalty means here, and what outcome the manager wants, whether a performance conversation or a termination.
Whether HR talks to the analyst depends on those answers. If the concern is loyalty alone, Miriam would coach the manager instead. The coaching starts with what conversations the manager has already had with her direct report.
Treat the Side Practice as a Signal
Rebecca has moonlighted, writing resumes in the early COVID years, and has managed people who did. In Rebecca’s experience, moonlighting “always has some sort of an unspoken need” behind it. Miriam turned that need into questions for the company, and honest answers depend on HR being seen as a partner.
If the reason is pay, HR can check its compensation strategy against the market. If it is skill, the analyst may be closing an AI skills gap the company left open. If it is a career move toward AI consulting, Miriam would consider bringing those skills in-house, because “every organization right now needs that.”
Write the Policy Around Customers and Company Resources
Miriam starts a policy with discovery too: why the company wants it, what it would do to a culture that prizes self-starters, and what it protects for the business. The line she called most important is the business itself. Side work in the company’s own space, or company tools and resources used to build a competing product, crosses it.
Rebecca counted customers and contacts gained through the job as company resources, with a simple rule: nobody should “double dip.” Side work for those customers is a conflict of interest in its plainest form. No law Rebecca knows of forbids it. Where the law is silent, a policy does the work, for instance by sending revenue from company customers to the company.
Miriam also checks what is legal in each jurisdiction, since many non-compete agreements are now illegal in a lot of them. She builds in a framework to revisit each policy, because keeping HR policies current means asking whether each one still needs to exist.
Where Employee Relations Fits
No policy was broken, and this is still an employee relations case. The damage is to a working relationship. Miriam read the manager’s concern as a trust problem. Rebecca saw the analyst’s side of it: a manager made an issue of a public post without raising it with the employee first. Miriam still counted it as a good sign that the manager came to HR, and she starts from the assumption that everyone has the best intentions.
The first decision becomes the precedent, so the record matters as much as the call. Teams that log each decision in HR case management software can see what they decided the last time a manager objected to lawful side work.
Frequently Asked Questions About Moonlighting in the AI Era
Not on a loyalty concern alone. Without a written rule, that concern stays subjective. The firmer grounds are a real performance problem, misuse of company-licensed tools under an acceptable use policy, or side work that reaches company customers.
It can be, if the acceptable use policy covers it. Many of those policies predate AI tools, so HR has to read what the policy actually says. Checking an employee’s AI history also raises privacy questions.
The manager. Miriam learns what prompted the concern, what the manager means by loyalty, and what outcome the manager wants. Whether the employee needs a conversation depends on those answers.
Start with the problem it solves. Weigh the culture it creates, its business impact, and what is legal in each jurisdiction. Draw the line at company customers and resources, and schedule a review.
Usually to meet a need the job does not. Rebecca and Miriam named three: pay that falls short of the market or of what life demands, skills the role does not teach, and an eventual career shift.
The Bottom Line for HR Leaders
Miriam Chan’s product manager framing is the useful one: discovery first, then the problem, and a rule only when the problem needs one. The scenario asked what the company could restrict. The better question is what it is trying to protect.
Her advice for anyone new to a case like this came from her own HR leader: “keep up with what is happening in the world.” HR people can stay inside their own expertise and miss what is changing around it. Staying well-rounded, Miriam finds, builds “empathy and understanding.”
See how AllVoices helps HR teams handle gray-area cases consistently, so the first decision is one the company can stand behind.
