“AI anxiety”: is worrying about AI taking your job affecting your health?

Worrying that AI will take your job is a form of perceived job insecurity, and perceived job insecurity is one of the better-studied occupational stressors there is. The evidence that it raises the risk of depressive symptoms, disturbs sleep and modestly increases cardiovascular risk comes from prospective cohorts and meta-analyses, not from surveys about AI. The AI-specific data are thinner: large surveys show that roughly two in five workers worry about AI making their duties obsolete, and that those who worry report more stress, but a survey cannot say which caused which. The evidence-aligned stance is to treat AI worry as a real stressor with known health effects, worth managing, rather than something to dismiss or to catastrophise.

How common the worry is

The most-cited figure comes from the American Psychological Association’s 2023 Work in America survey, run by The Harris Poll “among 2,515 employed adults nationwide between April 17 and April 27, 2023”. The APA reports that “Close to two out of every five workers (38%) reported worrying that AI might make some or all of their job duties obsolete in the future.”

The same survey found an association with stress: “Nearly two-thirds (64%) of those who reported being worried about AI possibly making some or all of their job duties obsolete also reported typically feeling tense or stressed during the workday, compared with over one-third (38%) of those who did not report being worried about AI.” Worried workers also more often reported feelings associated with burnout, such as “irritability or anger toward coworkers or customers (23% vs. 16% of those who do not worry about AI)”.

The APA is careful about what this means, and so should we be: the report notes that “there is no confirmation of the root cause of these feelings”. People who are stressed for other reasons may be more likely to worry about AI, and people in shakier jobs may be more likely to do both. A cross-sectional survey cannot untangle that. The APA’s 2024 survey repeated the pattern with slightly lower numbers (54% of AI-worried workers reported feeling tense or stressed during the workday, against 36% of those who were not worried).

The APA also notes that these worries “may be similar to what the media has labeled” AI anxiety. That is worth holding on to: it is a media label for a familiar thing, the belief that your job might not be there.

What job insecurity does to health

This is where the strong evidence lives: perceived job insecurity has been studied for decades, through recessions and restructurings, with designs that address causation better than a one-off poll.

Depression. A 2016 systematic review and meta-analysis of prospective observational studies, which follow people over time, pooled “20 cohort studies within 15 articles” and found that “job insecurity and unemployment were significantly related to a higher risk of depressive symptoms, with the odds ratio (OR) being modestly higher for job insecurity (1.29, 95% CI 1.06-1.57) than for unemployment (1.19, 95% CI 1.11-1.28)”. In these data, the fear of losing a job was at least as strongly associated with later depressive symptoms as actually losing one. The review also found “ORs for job insecurity were increased in longer exposure-outcome intervals (3-4 years)”, which fits a picture of chronic strain rather than a shock. An odds ratio of 1.29 is modest, and the review notes effects were strongest in unadjusted analyses, so treat it as a real but moderate risk factor.

A separate 2015 systematic review of the work environment and depressive symptoms, using GRADE ratings, listed job insecurity among factors with “limited evidence (grade two)”, alongside long working hours and low support; the strongest evidence was for lack of decision latitude and job strain. So job insecurity is on the list, but not at the top of it.

Sleep. In a 2018 analysis of 24,553 workers aged 25 to 65 across 31 European countries, “employment insecurity increased the odds of reporting insomnia or general sleep difficulties in the past 12 months. Each unit increase in employment insecurity elevated the odds of sleep disturbance by approximately 47%.” The finding held “across 27 of 31 European countries” after adjustment for age, education, contract type and household factors. This study is cross-sectional, so it shares the survey problem above, but the consistency across countries makes reverse causation an unlikely full explanation.

Heart disease. A 2013 meta-analysis in the BMJ combined individual data from 13 cohorts with four published studies, “up to 174 438 participants with a mean follow-up of 9.7 years and 1892 incident cases of coronary heart disease”. The age-adjusted relative risk for high versus low job insecurity was 1.32 (95% CI 1.09 to 1.59); after adjusting for socioeconomic position and risk factors, it fell to 1.19 (1.00 to 1.42), with the lower bound touching no effect. The authors’ own conclusion: “The modest association between perceived job insecurity and incident coronary heart disease is partly attributable to poorer socioeconomic circumstances and less favourable risk factor profiles among people with job insecurity.” In plain terms: a small excess risk, part of which is explained by insecure workers being poorer and less healthy to begin with.

None of these studies asked about AI. That is the point: the stressor is the perception that your work may disappear, whatever the cause.

Realistic appraisal versus clinical anxiety

Two things get blurred under “AI anxiety”.

The first is a realistic appraisal: your field is changing, some of your tasks are being automated, and you do not know what that means for you. That is not a disorder. It is accurate perception of uncertainty, and it can be useful because it motivates planning. The literature above describes what that uncertainty does if it goes on for years unaddressed.

The second is clinical anxiety. Generalised anxiety disorder is defined by the US National Library of Medicine as “excessive and difficult-to-control worry about a number of life situations”, where the “worry is accompanied by restlessness, fatigue, inability to concentrate, irritability, muscle tension, and/or sleep disturbance and lasts for at least 6 months”. The markers are excess relative to the situation, difficulty controlling it, physical symptoms and duration. If AI worry has become constant, has spread beyond work, and has affected sleep or concentration for months, that is a pattern a GP or psychologist should assess. A website cannot.

Most people sit between those poles: a realistic concern that occasionally tips into rumination. The research suggests leaving it there for years is not costless.

What the stress literature supports doing

The interventions with evidence act on the mechanism, not on AI.

Reduce the uncertainty you can. The APA’s own recommendation to employers is that “Open and honest communication from employers can help relieve employees’ anxieties about the unknown”. For an employee, the equivalent is asking direct questions about plans and, where answers are vague, treating the answer as unknown rather than filling the gap with the worst case. The 2016 meta-analysis found effects grew with longer exposure; shortening the period of not-knowing is itself protective.

Protect sleep specifically. Sleep disturbance is both an outcome of job insecurity in the European data and a symptom in the clinical definition of anxiety, and it is the lever most under your control: fixed wake times, no work email in bed, and persistent insomnia raised with a clinician rather than pushed through.

Convert worry into a plan where one exists. The difference between appraisal and rumination is whether the thinking ends in an action. Skills, savings and roles you can explore are actions. Re-reading layoff news is not.

Notice the amplifiers. Doom-scrolling AI coverage cues the worry loop the way the phone cues other compulsive habits; see is brain rot real. An AI assistant, tuned to agree, is a poor place to test whether your fears are proportionate.

Get help early if the pattern matches the clinical one. Anxiety that is persistent, uncontrollable and physical for months is a clinical matter. If you are in real distress, talk to a professional rather than to a website or a chatbot.

The honest summary

The specific fear is new. The physiology is not. Chronic perceived job insecurity is associated with a modestly higher risk of depressive symptoms (OR 1.29 in prospective cohorts), a substantially higher rate of reported sleep problems, and a small excess in coronary heart disease that partly reflects the circumstances of insecure workers rather than the insecurity itself. Whether AI worry specifically produces the same effects has not been studied prospectively; the survey data show association only. But the mechanism does not care what the threat is called. Manage it like the stressor it is. For how we weigh survey data against cohort studies, see how we research.

Common questions

How many workers are worried about AI taking their jobs?

In the American Psychological Association’s 2023 Work in America survey of 2,515 employed US adults, 38% reported worrying that AI might make some or all of their job duties obsolete. Among those who worried, 64% reported typically feeling tense or stressed during the workday, compared with 38% of those who did not worry. The survey cannot establish which causes which.

Does job insecurity cause depression?

Prospective studies suggest it raises the risk. A 2016 meta-analysis of 20 cohort studies found job insecurity was associated with later depressive symptoms with an odds ratio of 1.29 (95% CI 1.06 to 1.57), slightly higher than the odds ratio for unemployment itself. The effect is modest and partly attenuated by adjustment, and it grew with longer exposure of three to four years.

Can worrying about your job affect your sleep?

Yes, according to a 2018 analysis of 24,553 workers across 31 European countries, where each unit increase in employment insecurity raised the odds of reported insomnia or sleep difficulties by about 47%, consistently across 27 countries after adjustment. The data are cross-sectional, so causation cannot be proven, but sleep disturbance is also a core symptom in clinical definitions of anxiety.

Is AI anxiety a real anxiety disorder?

“AI anxiety” is a media label, not a diagnosis. Realistic concern about a changing field is not a disorder. Generalised anxiety disorder involves excessive, hard-to-control worry lasting at least six months with symptoms such as restlessness, poor concentration or sleep disturbance. If worry about AI has become constant, physical and long-lasting, that warrants assessment by a GP or psychologist.

Does job insecurity affect heart health?

Modestly. A 2013 BMJ meta-analysis of up to 174,438 people followed for an average of 9.7 years found a relative risk of coronary heart disease of 1.32 for high versus low job insecurity, falling to 1.19 after adjusting for socioeconomic and risk factors. The authors attribute part of the association to insecure workers having poorer circumstances and risk profiles.