Workplace safety in 2026 is increasingly a problem of categories OSHA does not regulate well. The Bureau of Labor Statistics' Survey of Occupational Injuries and Illnesses reports 2.6 million nonfatal injuries and illnesses in the U.S. private sector in 2023 — but that figure is widely understood to understate the true burden because it counts what employers report, and the largest contemporary occupational health categories (musculoskeletal disorders from screen work, psychosocial harm, heat-related illness for outdoor workers, indoor air quality, and the emerging algorithmic-management stress load) are systematically underreported. The unseen hazards are not exotic. They are the hazards that fall outside the legal definition of a reportable injury.
Fixing that requires updating the categories themselves, which is a slower-moving fight than the discourse about hard hats and forklifts suggests.
The musculoskeletal epidemic that the office invented
The Liberty Mutual Workplace Safety Index, an annual ranking of the most costly nonfatal workplace injuries, has for years had "overexertion" and "repetitive motion" injuries near the top of the list, with combined direct workers'-compensation costs in the multi-billion-dollar range annually. The CDC NIOSH (National Institute for Occupational Safety and Health) reports that work-related musculoskeletal disorders account for roughly 30% of nonfatal occupational injuries with days away from work. The largest single hidden category in office and computer-based work is upper-extremity repetitive strain — wrists, shoulders, neck — and the pandemic-era shift to home work made it worse, because home setups are typically less ergonomically appropriate than office setups, and few employers' return-to-office investments have addressed that.
OSHA's enforcement footing on ergonomic hazards is unusually weak. The Clinton-era ergonomic standard was repealed by Congress in 2001 and never replaced. Most ergonomic intervention in the U.S. now happens through employer voluntary action, workers' compensation pressure, and state-level rules (California has the most rigorous). The hazard is large, the cost is large, and the regulatory response is functionally absent — which is what "unseen" actually means in this context.
Heat: the fastest-growing hazard, regulated almost nowhere
The U.S. has no federal occupational heat standard. OSHA proposed a rule in 2024, with public comment running through early 2025; whether it survives the current administration is uncertain. Meanwhile, NIOSH and the CDC have documented year-on-year increases in heat-related occupational deaths and illnesses, concentrated among agricultural, construction, warehouse, and last-mile delivery workers. The Public Citizen / Public Health Watch reporting on warehouse heat illness — particularly the Amazon facilities documented during the 2022 and 2023 heat events — describes a hazard that is rising with climate change and largely uncontrolled by federal regulation.
This is the workplace-safety story climate change is writing in real time. Brookings Institution research on climate adaptation and labor estimates that heat-related labor productivity losses in the U.S. could reach the tens of billions of dollars annually by mid-century absent adaptation. The Climate Jobs → pillar explores the broader workforce implications. The narrower point here is that heat is a hazard whose burden is now mostly carried by the workers least protected by formal occupational-safety law.
The psychosocial category, finally legible
The ISO 45003 standard (Psychological Health and Safety at Work), published in 2021, is the first international standard treating psychosocial risk as a formal occupational-health category. It defines psychosocial hazards to include workload, control, work pace, role clarity, organizational support, recognition, change management, harassment, and work-life interface. The UK Health and Safety Executive's Management Standards for Work-Related Stress have operationalized similar categories in regulation since 2004, and the UK has measurable lower work-related stress prevalence than the U.S. on comparable survey instruments.
In the U.S., the EEOC and NIOSH are the closest analog, and the regulatory teeth are weaker. The American Psychological Association's annual Work in America survey finds roughly 60% of U.S. workers reporting that work-related stress affects their physical health, with chronic workplace stress now causally linked in the peer-reviewed literature to cardiovascular disease, diabetes, and depression. Christina Maslach's burnout work — most accessible in The Burnout Challenge (HBR Press, 2022) — frames this as a structural-design problem, not a personal-resilience one. The literature is increasingly clear that "workplace mental health" is occupational safety, and treating it under a separate regulatory regime understates the harm.
The hazard the technology itself is creating
The newer hazard category is algorithmic management. Warehouse productivity-monitoring software, ride-hail driver-pacing algorithms, food-delivery courier optimization, retail-shift micro-scheduling, and contact-center voice-stress monitoring all produce measurable physiological and psychological stress loads in workers, with documentation in the academic literature growing rapidly. Veena Dubal's work at UC Hastings on "algorithmic wage discrimination" and Alex Rosenblat's Uberland (UC Press, 2018) are early sources; Sarah Roberts's Behind the Screen (Yale, 2019) documents the parallel harms in content-moderation work.
The EU Platform Work Directive of 2024 begins to regulate algorithmic management directly. The U.S. has no equivalent, and OSHA is not currently constituted to address it. As more sectors adopt these tools, more of the daily occupational-stress load will fall outside the categories the regulatory system can see.
The agenda that closes the gap
Five interventions would close most of the visible-versus-actual hazard gap. First, a federal heat standard, which OSHA has proposed and which has strong evidence behind it. Second, restoration of a national ergonomic standard, given that musculoskeletal disorders are the largest single hidden injury category. Third, U.S. adoption of ISO 45003-style psychosocial-hazard reporting, even if voluntary, to begin building the data series we don't currently have. Fourth, transparency requirements for algorithmic management tools — the EU model is the template. Fifth, expansion of workers' compensation eligibility to cover chronic occupational-stress claims, which most state systems still treat skeptically.
None of these are exotic. They are the standard occupational-safety toolkit, applied to hazards the system has not yet learned to count. The "unseen" part of unseen hazards is mostly a regulatory choice, not a measurement problem.
"Workplace safety" still mostly means hard hats and forklifts in U.S. regulation. The actual hazard burden in 2026 is heat, repetitive strain, psychosocial load, and algorithmic management — and none of those are seriously regulated yet.
Updated May 21, 2026. This piece was substantively rewritten as part of NWLB's 2026 editorial refresh.
