The Real Reason Your Team Won't Share Their AI Workflows

Generative AI adoption inside companies rarely stalls because employees don't know how to use the tools. It stalls because they do not trust what will happen if they admit it.

That is the argument Eric Anicich and Jeslyn Brouwers make in their research on workplace AI transparency. The core problem is not a skills gap. Most knowledge workers, given access to a large language model, can figure out how to draft an email, summarize a document, or write a first pass at code. The problem is a trust gap.

An employee who uses ChatGPT to rewrite a client note, run a data analysis, or generate a presentation slide faces a real question: Does the company approve? Will the manager penalize the shortcut? Could the output get flagged as an ethics violation? When the rules around AI use are ambiguous, the rational move is to keep quiet.

That silence is expensive. The team loses the chance to learn what works. A junior analyst who figured out a faster way to query a database keeps it to themselves. A product manager who found a prompt pattern that cuts drafting time by half never mentions it. The organization's collective productivity gains stay locked inside individual inboxes.

Anicich and Brouwers argue that the fix is not a new policy document or a mandatory training session. It is creating conditions where sharing feels safe and worthwhile. That starts with removing ambiguity. If the company prohibits using AI on client-facing documents, say so explicitly. If the company permits it under certain conditions, write those conditions down in one place, not in a scattered set of Slack messages and all-hands comments.

The second condition is earned trust. Managers who react negatively to an employee's self-disclosure – even a mild