Pay Secrecy Is Making Your Best People Poachable
Originally Published in Inc.
New research reveals why the workers who look most loyal are building their exit case in real time.
U.S. workers send almost 3 million messages a day to ChatGPT, just about wages, compensation, and earnings. That’s from OpenAI’s own usage data, published in March 2026. Workers are using AI to figure out what a job pays before they apply, negotiate, or switch. Every day. At massive scale.
That story usually gets framed as a worker behavior trend. It’s not. It’s a leadership failure.
When your employees are using AI to benchmark their pay, they’re solving a problem you created. And the cost is almost certainly higher than you think.
The strategy that stopped working
For a long time, pay opacity served employers. Wage information was hard to get, socially awkward to ask about, and unevenly distributed through informal networks. If you knew what the market paid and your employees didn’t, that was a negotiating advantage and an indirect retention mechanism. Uncertainty kept people from knowing whether the grass was greener.
That friction is gone.
Workers can now get reliable national wage benchmarks in seconds. OpenAI’s research found that GPT-5.4 returns occupation-level wage estimates with a mean absolute percent error of 0.1 percent, nearly all falling within 10 percent of official benchmarks. Your employees are walking into their next performance review, or their next recruiter call, with better market data than most HR departments pull for salary conversations.
If your retention strategy depends on workers not knowing what they’re worth, you don’t have a retention strategy anymore.
The population you should actually be worried about
Wage searches on ChatGPT are not spread evenly. They concentrate in management, healthcare, and creative fields. They’re far less common in food service, office support, and production.
That pattern isn’t random. People search where pay is harder to ascertain and the stakes of getting it wrong are increased.
Those categories are your highest-value, hardest-to-replace roles. The workers doing the most compensation research are not your disengaged employees. They’re the ones with enough leverage to wonder if they’re underutilizing it, enough ambition to consider their options, and enough skill to actually have some leverage. These are your best people, in your most consequential positions.
The retention metric your dashboard is missing
Security is the dominant psychological need people bring to work. Not inspiration. Not career development. It’s the expectation that their income, role, and future are predictable and fair. We found that 95 percent of workers name financial security and job stability as fundamental expectations of employment. Only 47 percent say their organization actually meets that expectation.
That gap doesn’t stay quiet.
When employees don’t feel financially secure, something counterintuitive happens. On the surface, they look loyal. Stated turnover intention goes down. They don’t job search openly. Traditional retention metrics flag them as low risk.
What our research at Fractional Insights calls retention-tension is operating underneath. The same insecurity that makes people cling to the safety of the known also makes them far more receptive to an outside offer. Employees experiencing high insecurity report 45 percent higher poachability than those who feel secure, even while showing lower expressed intention to leave. For every 100 employees, about 23 more people are at flight risk than your intention-to-stay data would suggest.
They’re not planning to leave. They’re just not planning to stay.
The workers running ChatGPT salary searches are building the case for that moment. They’re learning their market value without you.
What to do about it
Pay transparency laws are spreading, and most HR teams treat them as a compliance problem to solve with minimum exposure. That framing misses the point.
When organizations communicate clearly about decisions that affect employees, our research shows people are 4.5 times more likely to have low workplace angst, and financial security is the dominant need driving that relationship.
Start with the information gap: Managers should be having proactive pay conversations, not reactive ones. If a direct report is wondering whether their compensation reflects their value, they should hear a credible answer from you, not from ChatGPT, and not from the recruiter who calls them on a Tuesday afternoon already knowing the answer.
Then make the conversation two-directional: Workers don’t just want a number. They want to understand the logic. What determines their pay, what would change it, and how does their role benchmark against the market? Opacity about how pay decisions get made is as damaging as opacity about the numbers themselves.
Pay particular attention to your high-uncertainty roles: Management. Healthcare. Creative. These are positions where compensation is most variable, most negotiable, and most actively researched by the people in them. They’re also the most expensive to backfill. If you don’t know how your pay compares to roles in the market, your employees probably do.
The last thing worth saying
Pay secrecy never protected your organization. It protected your negotiating position, which is not the same thing. The information advantage that used to favor employers now favors the employees with real options, and they’re using it.
You’ll figure this out proactively, or you’ll figure it out when your best person accepts an offer you didn’t know was on the table. The question worth sitting with: Do you know what your employees know about their own market value? And if not, who told them?
