Half the U.S. workforce is watching for new jobs and fewer than one in five are extremely satisfied with where they work. This isn’t disengagement in the old sense — it’s exit modeling: employees hedging against perceived instability by redirecting effort toward their own continuity rather than the company’s. It doesn’t show up cleanly on engagement dashboards, it’s a trust-deficit problem masquerading as a hiring-funnel problem, and the choice companies make now — AI to amplify human work or AI to cut headcount — decides who still has a workforce worth keeping in five years.

Exit modeling: your workforce is already repricing the employment covenant in real time.

Linda, a candidate, accepts an offer on Tuesday, signs the paperwork on Wednesday, and on Thursday tells the recruiter who placed her to keep looking for her. Jason, a new hire, starts on Monday, completes onboarding by Friday, and never quite tells his freelance clients he's done. Jackie, a manager, accepts a promotion, says the right things at the all-hands where the promotions are announced, and quietly stops referring friends while focusing more on networking outside of the company.

This is exit modeling. It begins before the start date and continues every day people are at work. It's heaviest in technology, but it's spreading everywhere.

Exit modeling is what people do when they no longer trust employers to act in good faith. They calculate, logically and emotionally, the risk that they'll be RIF'd, laid off, or quietly or unquietly exited at any moment, for any reason. They don't talk about it at work. You can be sure they're thinking about it.

Exit modeling is a framework for describing how workers hedge against perceived instability by prioritizing their own optionality. Gallup’s latest data show that 51% of U.S. employees are actively looking or watching for openings, and companies like Tesla illustrate a broader gap between CEO incentives and the people-and-culture goals firms may state elsewhere. Fewer than one in five are extremely satisfied with their employer as a place to work. This isn't a fringe sentiment. It's the dominant posture of the American workforce.

The reader who has been around long enough will say this isn't new. Employees have always hedged. People have always kept recruiters warm. That's true. What's new is the scale and the math. Gartner's January 2026 future of work analysis names "RIFs before reality" as the number one workforce trend for the year and reports that fewer than 1% of AI-attributed layoffs in 2025 were due to actual productivity gains. The employment deal, in Gartner's words, is being rewritten in real time.

Companies boasting record earnings while announcing workforce reductions isn't new. Doing it on the basis of an AI productivity bet that, by Gartner's measurement, is delivering transformative value in only one in fifty initiatives is new. The global workforce has noticed.

Exit modeling starts as unease, a gnawing feeling that all is not well, and quickly becomes an invisible form of disengagement that's far more costly than morale. Morale is a team feeling. Disengagement is personal. It seeps into consumer reputation, investor confidence, and the company's standing with its own future workforce. And ultimately into consumers, because we are each other's employees and customers. Left unchecked, a negative loop sets in. People exit model. The results start to show up in more obvious ways. Leadership calls for more automation to fill the gap. The cycle repeats.

Meta is mandating that US employees in certain departments allow AI to monitor their screens, keyboards, and mouse activity. A company is literally forcing its employees to be data entry widgets to improve AI agents at the kind of tasks people handle without thinking, like scheduling a meeting, booking a hotel, navigating a pulldown menu, recognizing that not all UX decisions are logical. Those workers are also training their replacements, and they know it. The same mandate is illegal in much of Europe. Whether US regulators come to see this as a human rights issue for the digital age, the workers being monitored already do.

Those workers are not going to be quietly quitting. Quiet quitting was a withdrawal of effort. Exit modeling is the opposite. It's a redirection of effort toward your own continuity rather than the company's.

Inside the behaviors

Candidates self-select for transactional fit. They stop evaluating culture, purpose, mission, growth narrative, and anything that isn't able to fit in a bank account. It would not be surprising if Meta's employees did everything they could to sabotage being data entry widgets. Users of Meta's platforms will face the same calculation. Spending is voting. Employees and candidates optimize for compensation, title, optionality, and most importantly, exit potential. They're not being shallow. They're pricing the relationship realistically. If the company is going to treat them as AI data entry widgets, and even the less ruthless versions, they'll treat the company as a paycheck machine. And if they feel they have nothing left to lose, then it's not a stretch to see smash the machine movements take root. Until then, the CHRO sees this as a quality-of-hire problem. Then a morale problem. Then engagement, attrition, and more. It's a trust-deficit problem masquerading as a hiring funnel problem.

Employees redirect discretionary energy toward exit readiness. Professional network maintenance, skill portability, side income, public profile building, recruiter relationships kept warm. This doesn't happen because they're disengaged in the old sense. They're engaged, just in their own continuity. CHROs see this as an engagement score declining. It's rational hedging behavior in a relationship where the company has retained all the optionality.

Neither behavior shows up cleanly in conventional metrics. With engagement surveys, exit modelers don't give honest feedback because they no longer believe it will make any difference. More likely, they’ll give glowing reviews, 5 out of 5 on every question. They don't work less. They work more, stockpiling skills, cramming learning resources, building public profiles. Individual productivity can look fine. Culture scores can look fine. The view from the dashboard is a bright sunny day. The view from the hearts and minds of the people inside is anything but.

The cost of getting this wrong isn't theoretical. Gallup's meta-analysis across 3.3 million employees in 347 organizations finds that business units in the top quartile of engagement see 23% higher profitability, 18% higher sales productivity, and 51% lower turnover than those in the bottom quartile. Engagement is the proximate driver of nearly every outcome an executive cares about. When exit modeling spreads, those drivers go with it.

Is AI to blame?

No, and yes.

The AI bet is not settled science, and it's not settled ROI math. Gartner reports that only one in fifty AI initiatives delivers transformative value. Among organizations piloting or deploying autonomous business capabilities, roughly 80% report workforce reductions. The reductions don't translate into return. Helen Poitevin, the Gartner analyst on the autonomous business survey, put it plainly: workforce reductions may create budget room, but they don't create return.

Yet the race is already underway. The themes are getting louder: if you're not an AI-native company, an AI operating system, you're at a competitive disadvantage. Says who? With what proof? Says the firms selling AI.

The fewer jobs people have, the smaller your customer base. A mutual self-destruction model is playing out, with no clear winner and very clear losers. Winning the AI race may mean losing the human race.

AI didn't cause exit modeling. It accelerated a repricing that was already underway. Companies are absorbing the cost of an unsettled productivity bet by transferring it to their workforce. The workforce has noticed.

The intangibles aren't intangible anymore

Companies used to rely on intangibles as differentiators and retention tools. Mission. Culture. Purpose. Growth narrative. They're fool's gold now in any industry where exit modeling has taken hold.

Employees and candidates in high-distrust industries, with technology first but spreading fast, will discount anything they can't take home at the end of the day. If your work is charged with building trust, engagement, culture, or reputation using strategies that worked years ago, you're in for a rough ride. The instruments you're relying on no longer measure what they used to measure, and they won't give you guidance for what's coming.

Why this is so hard to fix

The standard answer to engagement problems is to invest in managers. Gallup's research, across decades, finds that the manager accounts for roughly 70% of the variance in team-level engagement. That finding is real and it's reproducible. It's also incomplete for what's happening now.

Managers are downstream of the brand system. A great manager working under leadership the workforce has stopped trusting can buffer their team for a while. They can slow exit modeling on their own people. They can't reverse it. Eventually the contradictions between what the corporate brand says, what the employer brand promises, and what the C-suite does will break through the manager's own credibility. The team that was protected starts modeling exits faster than the rest of the organization, because the manager was the last reason to believe the employment trust account might still hold.

This is why fixing managers doesn't fix the problem at the current scale. The repricing isn't happening because managers are bad. It's happening because ownership of the employment covenant is fragmented in good companies and ignored in bad ones. TA wants to control what is said at the front door. HR wants to steer the experience inside. Comms wants to sway the external story. Internal comms wants to remind people how great things are. The C-suite wants to lay off everyone. Nobody wants to own the cracks, breaks, and fallout. Vigilance is what employees do when there's no clear steward of the covenant they're being asked to commit to.

It's also happening because corporate brand, employer brand, and culture are operating on different timelines and different assumptions. Corporate brand is bullish on AI productivity. Employer brand is still selling stability and growth. Culture is absorbing the layoff aftermath. Without integration, the contradictions become the lived experience employees are pricing.

This is why no single function, and no single manager, can fix it from their own seat. TA can't recruit its way out of a trust deficit. HR can't engage its way out of a credibility gap. Comms can't message its way out of contradictions the workforce is living through. Even the best manager can't out-coach a system that's actively undermining what they're asking their team to commit to. The repricing is happening at the level of the brand system, not at the level of any single touchpoint. It can only be seen, and addressed, at that altitude.

What good companies are doing

The "companies don't have a choice" argument is the standard defense for AI-driven layoffs. The data doesn't support it. In a separate Gartner survey of 321 customer service and support leaders, only 31% have implemented or are planning AI-driven workforce reductions through the first quarter of 2027. Eighty-five percent are expanding human agent responsibilities as AI reduces contact volume. Sixty-three percent are reducing frontline headcount gradually through attrition while reallocating capacity toward higher-value work.

The good companies are not pretending AI isn't happening. They're using the efficiency gains to expand what humans do, not to eliminate them. Eric Keller, the Gartner analyst on that survey, framed it as a strategic choice: organizations that only use AI to reduce costs risk missing the actual opportunity, which is combining AI efficiency with human judgment in roles customers value most.

This is what acting against exit modeling looks like in practice. Not a trust campaign. Not a values refresh. A demonstrated, ongoing commitment to expanding human work as AI capability grows, communicated through what the company actually does rather than what it says. When the system is coherent, great managers become the multiplier the engagement research has always said they could be. When it isn't, no amount of manager training compensates.

Bad companies will not make this choice. They'll be the first to the bottom. The incentives at the top tell the story. Tesla's trillion-dollar pay plan was tied to financial and operational milestones, not to employee engagement or community well-being. That creates a gap between the company's stated people-and-culture goals and the incentives on its CEO. Consumers and employees, who are often the same people, will price that into how they vote. For good companies, the work is to give people reasons to trust, even if they're exit modeling. This is the counterintuitive part. When fear-based decision making doesn't produce the worst case, people don't release the fear. They extend it forward. The only way out is to give them concrete reasons to recalibrate, again and again, over time. You're not going to talk anyone out of exit modeling. You're going to give them reasons to slow it down, and then to question it, and eventually to set it aside and invest that energy in continuing to be a good, strong company.

That means decoupling promises that can't be kept from ones that can. It means making sure what one part of the organization says doesn't undermine what another part does. It means treating the employment covenant as a coherent system with a clear steward, rather than as a series of handoffs between functions that don't work well with each other.

On AI specifically, it's not hard to see employees voting as consumers, especially those displaced by AI layoffs. They'll vote against companies that treat AI as the workforce, and toward companies that treat AI as the tools their workforce uses. The long-held adage that saving money isn't the same as making money is about to become an executive-level conversation again.

You can't tell someone to trust you. You can show them, in how the company actually operates, that trust is warranted.

The recognition

Your workforce is already operating on a repriced relationship, even if your company had nothing to do with it. The metrics that would tell you this aren't the ones being tracked. Engagement looks ok enough. Productivity looks ok enough. Quality of hire looks like a media channel problem.

None of the above. Exit modeling is happening across your organization right now, in people who haven't said a word about it and won't.

Do you acknowledge the repricing, or pretend it's not happening because it's hard to quantify? Gartner has named the trend. Gallup has measured the watching-for-openings posture at 51% of the workforce. The data says the AI productivity bet isn't paying off and the workforce knows it. Companies are sorting themselves into two groups: the ones using AI to amplify human work, and the ones using it to reduce headcount with no measurable return. The second group is racing to the bottom whether they know it or not.

The first group is the one that still has a workforce worth keeping in five years, and the one that will earn the votes to continue. How will you vote, as an employer, employee, and consumer?