A founder praises AI at an all-hands meeting, then jokes that a manager used it to draft a customer email. A CEO urges employees to experiment, then applauds someone for producing a report “without AI.” Those small signals teach employees that AI use may receive formal encouragement while carrying social risk.
Microsoft and LinkedIn found that this dynamic already produces widespread shadow AI, with 78% of AI users bringing their own tools to work, while 52% hesitate to admit using it for their most important tasks and 53% worry that doing so makes them look replaceable.
Dr Gleb Tsipursky, CEO of Disaster Avoidance Experts and author of The Psychology of AI Adoption at Work: From Resistance to Results, devoted a section of his new book to shame and hidden AI use. He argues that leaders often create shadow AI through the social signals they send. “Shame turns AI use into a secret,” writes Dr Tsipursky. He describes how leaders can break that secrecy loop through four practical changes.
1. Define acceptable use before demanding disclosure
Many companies tell employees to use AI responsibly without explaining what responsible use means. Workers then have to guess which tools they may use, what information they may enter, when disclosure is required, and who owns the final decision. The NIST AI risk management framework gives leaders a useful foundation for turning broad principles into explicit practices for managing AI-related risk.
Dr Tsipursky says leaders should divide work into four categories: approved uses, approved uses that require disclosure, restricted uses that require permission, and prohibited uses. IBM Research found that more granular AI disclosure can reduce perceived stigma when employees say how AI contributed, confirm that they followed policy, and state that a human reviewed the result.
A useful disclosure, according to Dr Tsipursky, might read: “AI generated the first draft. I checked the facts, revised the reasoning, and approved the final version.” That gives colleagues meaningful information about accountability without forcing employees to defend every use of the tool.
2. Model transparent disclosure from the top
The social risk around workplace AI reaches well beyond a few reluctant adopters. Pew Research Center found that 52% of US workers feel worried about future AI use in the workplace, and 33% feel overwhelmed. Harvard Business Review has also described an AI penalty that can attach to workers when others know they used AI.
Leaders can either reinforce that stigma or weaken it. A founder who quietly uses AI for speeches while mocking an employee for using it on a memo teaches everyone that disclosure creates personal risk. Gallup found that employees who strongly agree they have manager support for AI use are twice as likely to use AI frequently.
Dr Tsipursky advises that executives and managers should regularly explain where they used AI, where the output fell short, what they changed, and what judgment they applied before accepting the result. That changes the meaning of disclosure from “I could not do this myself” to “I used a tool and remained accountable for the outcome.”
3. Reward judgment instead of performative self-sufficiency
Some workplaces still reward the appearance of solitary effort. Employees gain status by looking as though every sentence, analysis, and idea emerged without assistance. That incentive encourages people to conceal both the tool and the mistakes it produced. Research from BetterUp Labs and the Stanford Social Media Lab found that AI workslop reached 40% of surveyed US desk workers in a one-month period, with each incident taking an average of two hours to resolve.
Dr Tsipursky says managers should evaluate the quality of the final work, the employee’s reasoning, the verification process, and the business outcome. An employee who catches an invented fact before it reaches a customer demonstrates more value than someone who quietly passes along a polished but unreliable answer.
4. Make it safe to report when AI fails
A shame-based culture turns small AI errors into hidden liabilities. Employees who expect ridicule or punishment have an incentive to delay reporting inaccurate summaries, exposed data, weak customer communications, or biased recommendations. SHRM found that workers report higher engagement and stronger commitment when organisations take an open approach to AI integration.
Dr Tsipursky captures the leadership test directly: “Each AI slip or ‘hallucination’ tests leadership’s ability to build a learning culture rather than assign blame.” A 2026 two-wave survey of 635 marketing employees found that disclosure silence emerged through multiple pathways involving fear of negative evaluation, AI anxiety, creativity threat, job insecurity, low trust in management, weak psychological safety, and unclear AI policy.
Leaders should create a no-blame process for reporting AI failures and near-misses, and then train managers to respond with questions rather than accusations. Ask what the employee was trying to accomplish, what the tool produced, what the employee noticed, and what safeguard could prevent a repeat. That approach brings problems into view while there is still time to contain them.
A company that encourages AI while shaming employees for using it will get concealment instead of control. Clear rules, visible disclosure from leaders, rewards for sound judgment, and safe reporting make AI adoption easier to govern because employees have fewer reasons to hide how the work actually gets done. – Inc./Tribune News Service
