Managers trust employees less when they know AI was involved in their work, even when the work itself isn’t part of the judgment. According to Futurism AI, a new study in the journal The Leadership Quarterly found that supervisors’ trust in an employee’s report fell from 68 percent to just 24 percent once they knew the employee had used AI. That’s a big penalty for using a tool many of those same companies are pushing their staff to adopt.
The research team was led by Samuel Meyers, a PhD student at Hebrew University of Jerusalem. Their results add to the growing evidence that the workplace AI push has been messier than promised.
🔬 What the Researchers Did
The team ran a series of experiments with one simple setup. Supervisors had to decide whether to trust an employee’s report, and they were told whether that employee had used AI or not.
What matters here is that the supervisors weren’t basing their judgment on the quality of the work. The only thing that changed was whether AI was in the picture. That design isolates the bias itself, so whatever shifted came from the AI label alone.
📊 The Numbers
| Condition | Chance of earning supervisor trust |
|---|---|
| Employee did not use AI | 68% |
| Employee used AI | 24% |
That’s a drop of 44 percentage points. In relative terms, an AI-assisted employee was roughly a third as likely to be trusted. Futurism AI reports that this trust gap actively hurts employees’ chances of promotion, and that it shows up even at workplaces that still encourage AI use.
“AI is changing not only how we work, but also how we judge the people we work with,” coauthor Yochanan Bigman, a senior lecturer at Hebrew University of Jerusalem, said in a statement. “An employee may use AI to produce better work more efficiently yet still lose their manager’s trust simply because AI was involved.”
🏢 Why This Hits Now
The study comes as executives are rethinking their own AI mandates. Futurism AI points to several signs of a backlash:
- Workslop fatigue: A tech recruiting CEO told Business Insider she banned AI-generated Slack messages for internal communication after being flooded with them.
- Open frustration: One anonymous CEO threatened to “fire the next person” who sent unedited ChatGPT output, as Inc reported earlier this year.
- Rising costs: AI agents and coding tools have pushed spending up so much that some companies have had to cut back.
- No clear payoff: Polls show most CEOs haven’t seen productivity gains from AI.
What stands out to me is the contradiction. Leadership tells people to use AI, then trusts them less when they do. Employees end up in a lose-lose spot: follow the mandate and take a credibility hit, or ignore it and risk looking out of step.
⚠️ The Shadow AI Risk
Meyers and his colleagues warn that this tension could lead to “shadow AI,” meaning undisclosed AI use at work. If being open about AI costs you trust, people will simply stop telling anyone.
That’s a real problem for companies. Hidden AI use makes it harder to check quality, protect sensitive data, or learn which workflows actually benefit from the tools. It could also hurt the workers themselves if their undisclosed use comes out later.
🛠️ What You Can Do With This
For managers:
- Check your own reaction. If you trust a report less just because AI touched it, judge the output instead of the process.
- Set clear disclosure rules so employees don’t have to guess what’s acceptable.
- Separate “used AI” from “didn’t review the work.” The second one is the real issue with workslop.
For employees:
- Never send raw AI output to your boss. Edit it, check it, and make it yours.
- When you disclose AI use, explain what you checked and decided yourself.
- Treat AI as a drafting tool, not a stand-in for your own judgment.
🔍 Limitations Worth Noting
The experiments deliberately took work quality out of the equation. That’s useful for measuring bias, but real managers usually see the work too, and a strong report could shrink the gap. The coverage also doesn’t break down how trust shifts across industries, seniority levels, or types of tasks. Those are open questions for future research.
Bigman’s main point still holds: “organizations need to recognize that adopting AI is not just a technological challenge, it is also a psychological and managerial one.” Companies that skip that part may find their AI rollout doing quiet damage to trust inside their own teams. You can find the full study details in Futurism AI’s original report.