Your Coworkers Use ChatGPT First, Worry About Privacy Later
Your office may be leaking secrets to AI without anyone noticing.
Researchers at a German university surveyed 224 employees across education, healthcare, IT, and other fields to see how they handle ChatGPT's privacy risks in their daily work. The study, reported on arXiv in August 2026, found that people tend to "chat first and worry later" — they turn to the AI for help writing, summarizing, and solving problems without thinking about what data they're giving away.
The big surprise: even when companies have official policies about using generative AI, employee knowledge about privacy risks stays weak. In other words, reading a policy isn't the same as truly understanding what ChatGPT does with the text you paste in. The researchers measured something called "privacy-related proficiency" and found it's generally low across all workplaces.
The study also showed that people who are already worried about privacy tend to use ChatGPT less often and for fewer tasks. This effect is strongest in companies that have no AI policies at all — employees there are both less cautious and less informed. This creates an uncomfortable situation: the people who need protection the most are the ones least likely to get it.
So what can you do? If you're an employee, treat anything you type into ChatGPT as something a stranger could read. If you're a manager, remember that a one-page policy isn't training. The researchers suggest that practical, hands-on education about AI privacy should be as routine as learning any new software.
- Most employees use ChatGPT at work without fully understanding what happens to the data they paste in.
- Company policies about AI use barely improve workers' actual privacy know-how.
- People with higher privacy anxiety cut back on using ChatGPT, especially when no company guidelines exist.
Why It Matters
If you use AI at work, this could mean your company secrets are already in someone else's hands.