Answers are grounded in this paper and cite their sources.
Employers are adopting software that tracks activity, scores behaviour and flags deviations, and surveys suggest that half of U.S. workers already believe they are monitored. This study asks what such monitoring is associated with in the places where trust and psychological safety are formed. It assembles the published results of two Harris Poll surveys for the American Psychological Association (2,515 workers in 2023 and 2,027 in 2024), Gallup’s September 2026 study of trust in leaders’ use of employee data, and Gallup’s 2026 workforce surveys on AI, and analyses them with explicit effect-size formulas: risk ratios, Cohen’s h, random-effects pooling, population attributable fractions and implied-rate bounds. Across ten published contrasts, monitored workers fared worse on every outcome, with a pooled effect size of h = 0.36 (95% interval 0.22 to 0.49 after allowing for correlated outcomes) and substantial heterogeneity (I-squared 70 percent). They were nearly twice as likely to plan to leave within a year (39 versus 21 percent; risk ratio 1.86, 95% interval 1.56 to 2.21), which would correspond to about 30 percent of such intentions if the association were causal. Trust in leaders’ use of employee data is low (20 percent a great deal) and steeply related to concern (13 to 65 percent across trust levels); only 19 percent of employees have ever been told what is collected, and conversations raise both trust and concern. The evidence is cross-sectional, self-reported and partly drawn from press summaries, so the findings are associations. Purpose - This study examines how AI-enabled workplace monitoring relates to psychological safety, trust and strain, and what published evidence implies about the conditions under which monitoring is tolerated. Design/methodology/approach - The paper reanalyses published survey contrasts from the American Psychological Association and Gallup using effect-size formulas, random-effects pooling with a correlation adjustment, replication tests across survey waves, attributable-fraction calculations and implied-rate bounds, with sensitivity checks for assumed group sizes and design effects. Findings - Monitoring is associated with higher strain, lower perceived respect and trust, and exit intentions; effects are heterogeneous but consistently positive. Trust in data use is closely tied to engagement and concern, and transparency raises trust and concern together. Research limitations/implications - Group sizes and design effects are approximated, some sources are secondary, and the data are cross-sectional self-reports with no causal identification or direct measure of psychological safety in the monitoring contrasts. Practical implications - Employers should specify what is monitored and why, involve workers in setting limits, keep humans in decisions that affect people, and expect transparency to raise questions before it builds trust. Social implications - Monitoring that workers read as distrust may carry costs in strain and turnover that are not visible in productivity dashboards. Originality/value - The paper combines two decades of theory on trust and psychological safety with the most recent U.S. survey evidence on AI-era monitoring and quantifies the size, heterogeneity and sensitivity of the reported associations.
No takes yet. Share an insight, caveat, or question.
Sarfaraz Abdullah (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: