- AI was cited in 101,743 US job cuts in the first half of 2026, nearly double the total for all of 2025, and it has been the leading stated reason for workforce reductions since March.
- In a June 2026 survey of 1,000 US employees, 36% said going above and beyond would not save their job, so they saw little reason to give extra effort.
- This is not laziness. Steven Maier and Martin Seligman's 2016 reanalysis of learned helplessness shows that disengagement under uncontrollable threat is the nervous system's default, and effort is the thing that has to be actively generated.
- Johannes Siegrist's effort-reward imbalance model predicted this outcome in 1996: high effort with low or uncertain reward produces withdrawal and measurable health cost.
- Andy Challenger's observation that firms are cutting headcount to fund AI spending, not to replace specific roles, means the threat is unpredictable, which is precisely the condition that produces helplessness rather than motivation.
One research-backed insight per week on stress and nervous system regulation — free.
No spam. Unsubscribe anytime.
In This Article
Something has changed in how people are responding to job threat, and it is the opposite of what most executives expected. Faced with the most visible wave of AI-attributed job cuts yet recorded, workers are not competing harder. They are stepping back.
The numbers are stark. AI was cited as the reason for 101,743 US job cuts in the first six months of 2026, nearly double the 54,836 attributed to AI in all of 2025, and it has been the most commonly stated reason for workforce reductions every month since March. And in a June 2026 survey of 1,000 US employees, 36% said that going above and beyond would not save their job, so they saw little reason to give extra effort.
Read that second figure again. More than a third of the workforce has concluded that performance and survival are no longer coupled. This is being called quiet cracking, and treating it as a motivation problem is going to fail, because it is not one. It is a textbook response to uncontrollability, and the psychology has been documented for close to sixty years.
What Makes This Different From Quiet Quitting?
Quiet quitting, the phrase that dominated 2022, described a choice. Workers were setting boundaries, declining unpaid extra, deciding that the job would get contracted hours and nothing more. It was framed, correctly or not, as an act of agency.
Quiet cracking is not an act of agency. It is closer to something giving way under load.
The distinction matters because it changes the intervention. You can negotiate with a boundary. You cannot negotiate with a nervous system that has concluded its efforts do not affect its outcomes. One is a position. The other is a state.
And the state is being produced by a specific feature of this particular wave of cuts: the threat is not legible. Andy Challenger has pointed out that many of the reductions being attributed to AI are not cases of a model doing a named person's job. Companies are reallocating budget toward AI capability and cutting headcount to fund it. The consequence is that vulnerability is no longer concentrated in roles that are obviously automatable. It is distributed almost arbitrarily across the org chart.
An arbitrary threat is a very different psychological object from a legible one. If you know the risk sits with a specific skill, you can respond to it. If the risk is that your department's budget line becomes an inference bill, there is nothing to respond to.
Why Uncontrollable Threat Produces Withdrawal, Not Effort
This is the part organisations keep getting backwards, and the research is unusually clear.
The classic account comes from Martin Seligman and Steven Maier's experiments beginning in 1967, which established what became known as learned helplessness: subjects exposed to aversive conditions they could not influence subsequently failed to act even when action would have helped. For decades this was read as learning to be helpless.
Maier and Seligman revised that interpretation in a 2016 paper in Psychological Review, and the revision is the interesting part. Drawing on the neuroscience that had accumulated in the intervening fifty years, they argued the original reading was backwards. Passivity under prolonged aversive stimulation is not learned. It is the default response of the relevant brain circuitry. What actually gets learned, when the situation permits it, is control, and that learning is what inhibits the default passivity.
The implication for a workforce is direct. Disengagement is not something employees acquire through bad attitude. It is what happens when the conditions that normally teach control are removed. Effort is the state that requires maintenance. Withdrawal is what the system does when nothing is holding it up.
So when a third of a workforce reports that effort will not change their outcome, you are not looking at a cultural failure. You are looking at the exact input condition the research says produces exactly this output.
The Model That Predicted This in 1996
Johannes Siegrist's effort-reward imbalance model, introduced in 1996 in the Journal of Occupational Health Psychology, described the mechanism before anyone had heard of a large language model.
Siegrist's argument was that occupational strain is not primarily a function of how much work you do. It is a function of the ratio between what you put in and what comes back, where the return side includes money, esteem, career progression, and job security. When high effort meets low or uncertain reward over a sustained period, the result is not just unhappiness. Siegrist's model, and the large body of evidence that has since tested it, links sustained imbalance to elevated cardiovascular risk and depressive symptoms.
Job insecurity is a reward-side variable. A worker whose security has been removed has had their side of the ledger cut, regardless of whether their salary changed. And the rational response to a worsened exchange rate is to reduce what you supply.
Mats Sverke, Johnny Hellgren and Katharina Näswall's 2002 meta-analysis, also in the Journal of Occupational Health Psychology, pooled the job insecurity literature and found consistent negative associations with job satisfaction, organisational commitment, trust, and both psychological and physical health. Notably, the effects of insecurity often rivalled those of actual job loss. The anticipation does much of the damage.
That is worth sitting with. The 101,743 people whose jobs were cut are the visible cost. The much larger group watching, and recalculating, is where most of the aggregate damage is accruing.
What This Does to the Body
The cultural conversation about layoff anxiety tends to stay in the register of morale. That undersells it.
Sustained, unpredictable threat is a physiological condition, not just an emotional one. The stress response was built for threats that resolve, and its costs are manageable when activation is followed by recovery. A threat that neither materialises nor disappears removes the recovery half. The system stays partially mobilised for months.
We have written about what that accumulation does in What Is Nervous System Debt? and about how it eventually surfaces as physical symptoms in The Somatic Veto. The short version is that chronic low-grade activation is not a milder version of acute stress. It is a different and in some ways more corrosive pattern, because it never triggers the response that acute crisis does. Nobody takes leave for a threat that has not happened yet.
It also reliably wrecks sleep, and does so in a specific way: the mind will not stop running scenarios at the moment the day's distractions are removed. If you are lying awake rehearsing a redundancy conversation that may never occur, that is not a sleep hygiene problem, and it is covered in Sleep Anxiety and the Bedtime Threat.
Why "Resilience Training" Will Make This Worse
The predictable organisational response to a disengagement statistic is a wellbeing initiative. Resilience workshops. A mindfulness app licence. Perhaps a webinar on managing uncertainty.
There is a specific reason this backfires here, and it is not the usual complaint that such programmes are superficial.
Offering an individual-level intervention for a structurally generated problem communicates something quite precise to the workforce: that the organisation has located the fault inside them. When the actual input is an uncontrollable threat created by budget decisions taken several levels above, a course on personal resilience reads as an instruction to adapt to conditions rather than an acknowledgement of them. It adds a layer of implied personal failure on top of an already depleted system.
This is the pattern we described in The Self-Care Industry and in When Systems Remove Recovery. The interventions are not useless in themselves. They are misapplied, and misapplication has a cost of its own.
The things that would actually move the number are unglamorous and expensive: clarity about what is happening and when, honesty about which functions are exposed, decision transparency, and restoration of some genuine control over how work gets done. All of those operate on the controllability variable, which is the one the research says matters.
What to Do If You Are the Worker in This
Most readers of this are not the person setting the redundancy policy. So the more useful question is what a reasonable response looks like from inside it.
Stop treating your disengagement as a character defect. It is a proportionate response to a poor exchange rate. Recognising it as a calculation rather than a failing removes the second, self-directed layer of stress that tends to sit on top of the first.
Separate the threat you can act on from the threat you cannot. Skills, network, and financial runway are controllable. Whether your employer reallocates its budget toward compute is not. Effort spent on the first category returns something. Effort spent worrying about the second is the exact input that produces helplessness, because it is effort aimed at an uncontrollable variable.
Protect the recovery, specifically. Under sustained threat, recovery is the first thing to go and the thing that most determines how long you can hold. That means real psychological detachment rather than merely being off the clock, which we break down in Which Type of Rest Do You Actually Need?.
Reinvest the withheld effort somewhere it compounds. The discretionary energy you have stopped giving the job does not have to evaporate. Mastery in a domain you control is one of the few reliable generators of the sense of agency that the situation is stripping out.
Watch for the physical markers rather than the emotional ones. Emotional self-report is unreliable under chronic stress because the baseline shifts. Sleep, appetite, resting heart rate, and how quickly you become irritable are more honest instruments. The Signs of Nervous System Burnout covers what to look for.
The Uncomfortable Conclusion
Quiet cracking is being reported as a workforce problem. It is more accurately a workforce symptom, and the underlying condition is that a large number of organisations have, over about eighteen months, severed the link between individual effort and individual outcome, and then expressed surprise that individual effort declined.
The employees have not become worse. The exchange has. And the research going back to 1967 says that when you remove controllability, withdrawal is not what people choose. It is what is left when the thing that was holding effort up gets taken away.
You are not broken for feeling this. You are responding correctly to information about your situation. The question worth asking is not how to force the effort back, but what would have to change for it to become worth giving again.
Get one essay a week on the science of rest and the cost of optimization culture. Subscribe to the newsletter. Free, one essay a week, no upsell.
Related reading: AI Anxiety and the Nervous System · When Systems Remove Recovery · What Is Nervous System Debt? · Hustle Culture Is a Structural Problem
The research is actionable. So is the newsletter. One insight per week, no fluff.
No spam. Unsubscribe anytime.
Frequently Asked Questions
What is quiet cracking?
Quiet cracking describes a workforce state in which employees remain employed and outwardly compliant while their discretionary effort collapses. It is distinct from quiet quitting, which was framed as a deliberate boundary-setting choice. Quiet cracking is closer to attrition of capacity than to a decision. Workers still attend, still deliver the minimum, and have stopped generating the extra initiative, ideas, and care that organisations depend on but rarely pay for directly. The 2026 version is being driven substantially by AI-related job insecurity rather than by workload alone.
Why are workers disengaging instead of working harder during AI layoffs?
Because the threat is not legibly connected to performance. Andy Challenger of Challenger, Gray and Christmas has noted that many AI-attributed cuts are not cases of software replacing a specific person's work, but of companies reducing headcount to free budget for AI investment. That makes the risk feel arbitrary. Decades of research on controllability, beginning with Martin Seligman and Steven Maier's work in the late 1960s, shows that when outcomes appear uncoupled from effort, organisms reduce effort. This is a predictable response to unpredictability, not a character failure.
How many jobs have been cut because of AI in 2026?
AI was cited as the reason for 101,743 US job cuts in 2026 through the end of June, according to figures reported that month. That is nearly double the 54,836 attributed to AI across the whole of 2025. AI has been the single most commonly stated reason for workforce reductions in each month from March through June 2026. It is worth reading these numbers carefully: they record the reason employers gave, which is not identical to the reason the job disappeared.
What is effort-reward imbalance?
Johannes Siegrist introduced the effort-reward imbalance model in 1996, publishing in the Journal of Occupational Health Psychology. It holds that strain arises not from workload in isolation but from a mismatch between the effort a person puts in and the rewards they receive back, where rewards include money, esteem, security, and career prospects. Job insecurity sits squarely inside the reward side of that equation. Siegrist's model has since accumulated substantial evidence linking sustained imbalance to cardiovascular risk and depressive symptoms, which makes disengagement look less like a motivation problem and more like a protective response.
Is disengagement bad for the worker or protective?
Both, on different timescales. In the short term, withdrawing discretionary effort under conditions you cannot control is genuinely protective. It reduces the gap between what you spend and what you get back, and it limits the damage of investing heavily in something that may be taken away regardless. Over longer periods it carries its own cost, because meaningful work is one of the more reliable sources of psychological resource, and disengagement forfeits that. The honest framing is that it is a rational trade made under bad conditions, not a flaw to be coached away.