Employees hide their AI workflows, and companies lose know-how

Harvard Business Review published an article by Eric Anicich and Jeslyn Brouwers on June 10, 2026 about why employees do not share their ways of working with AI. The authors describe a situation that will become increasingly common in companies: people create useful workflows, prompt templates or small shortcuts, but keep them to themselves. Not because they lack a company tool. Often because they do not know how their AI use will be judged by a manager, colleagues or HR. HBR gives the example of a 31-year-old doctor who used the approved DoximityGPT tool several times a day, with compliance with US HIPAA rules. He created his own template, which he said produces astoundingly good results. Colleagues were struggling with the same tool. Yet he hesitated to share the workflow. This is the central problem: the company has both the tool and a useful practice, but the knowledge remains hidden with one individual. The authors link this behaviour to trust and learning culture. If an employee feels that admitting AI use may raise suspicion of laziness, rule-breaking or job risk, the employee will stay silent. The company then loses two things at once. It cannot see real cases where AI works, and it cannot detect risky use that should be corrected. Instead of managed learning, a grey zone appears. The business impact is concrete. One employee may save an hour a week with AI, but if the workflow is not shared, the saving does not spread. Another may use AI on sensitive data, but if nobody knows, the risk remains out of sight. From a leadership perspective, it is therefore a mistake to treat AI only as a question of licences and security policy. People must also be willing to show what they really do. The practical answer is not a hunt for employees. Companies need a safe space for sharing workflows, clear boundaries for prohibited data and quick recognition of good examples. If someone creates a template that improves team work, it should be seen as process improvement, not as a trick to avoid work. AI know-how must move from personal advantage to company capability. Otherwise businesses will pay for tools while leaving most learning locked inside individuals’ heads. Terms: An AI workflow is a repeatable process for using AI on a specific work task. HIPAA is a US law protecting patients’ health information. Grey AI use means unofficial AI work outside clearly described company rules.

Employees hide their AI workflows, and companies lose know-how

The full article is available to users registered at Hard Skills