Employee Privacy and Human Review in AI Workplace Monitoring in Bangladesh

Artificial intelligence can expand the ability of employers to collect and analyse information about work. Whether such monitoring is acceptable depends on both informational safeguards and the procedures through which monitoring data affect employees. This study examines expectations of notice, proportionate purpose, data disclosure and human review in Bangladesh. It analyses 385 adult respondents who reported awareness of AI, using item-level distributions, paired endorsement comparisons and an exploratory proportional-odds model. The sample includes 238 law students and is not presented as representative of employed workers or the national workforce. Notice that AI monitors workplace activity was endorsed by 292 respondents (75.8%), while review before serious employment consequences was endorsed by 205 (53.2%). Their within-person difference was 22.6 percentage points (95% bootstrap interval 16.4-28.8; Holm-adjusted p < .001). Sixty-one respondents (15.8%) endorsed all four procedural safeguards. Privacy safeguards were said to affect acceptance of monitoring by 301 respondents (78.2%). In the adjusted model, the notice expectation showed a pointwise odds ratio of 1.35 (95% confidence interval 1.04-1.75), but its Holm-adjusted p-value was .102; the four focal associations therefore provide limited evidence after multiplicity adjustment. These findings describe distinctions between stated expectations, not the effect of actual monitoring, privacy violations or psychological-contract breach. Integrating legal safeguards with job demands-resources and psychological-contract perspectives, the article proposes a sequence connecting notice, purpose limitation, usable information and meaningful review. Longitudinal and experimental studies of employees and implemented systems are needed to examine whether these safeguards change perceived privacy and workplace outcomes.