Some principles for understanding AI hazards in the workplace

John P. Sadowski, Ph.D.

Read the full paper

Online version from The Synergist (requires AIHA login) | Purchase The Synergist issue

As systems enabled by AI become increasingly introduced into the workplace, OSH practitioners and researchers must be prepared to consider their effects on worker health and safety. While AI may at first seem radically different from other potentially hazardous aspects of the workplace, it can be understood using established OSH principles. A framework has been devised that identifies several core characteristics of algorithmic systems and links them to existing, well-known categories of occupational hazards and controls. Key points of the framework include:

1. Scope. The term "trained algorithm," meaning use of a training data set to create an algorithm for a system's output or behavior, is a better conceptualization than "artificial intelligence" because it is a practical description of a specific process.
2. Identification. Algorithms must be distinguished from the physical platforms they may be embedded in, but platforms can help identify trained algorithms being used in a workplace.
3. Hazards. Because algorithms are software with no physical substance, they cannot directly create any "tangible" physical, chemical, or biological hazards, but they do alter the risk profile of physical platforms or substances they control or interact with. By contrast, they can directly create psychosocial hazards by changing work organization, skills, and relationships.
4. Exposure assessment. Established and familiar exposure assessment methods can still be used even without AI subject matter expertise, which can be considered "hazard-side" assessment. Assessing the impact of specific system characteristics of algorithms, which can be considered "system-side" assessment, will enhance the assessment.
5. Controls. There is a distinction between "prevention through work design" hazard controls that can be carried out within the end-user's organization, and "prevention through software design" controls that must be applied by the software developers, possibly at a separate vendor firm.

These priciples and the accompanying framework are intended to enable a science of "algorithmic hygiene" that explicitly links algorithmic system characteristics to health and safety outcomes. Ultimately, this would guide research questions and provide a scientific basis for practical, actionable guidance for individual OSH professionals, end-users, developers, and policymakers.

News:

• April 2026: I gave an invited webinar for the Worker Training Program of the U.S. National Institute of Environmental Health Sciences in preparation for a workshop on AI and the Future of Worker Safety and Health Training.
• January 2026: This work was featured in a NIOSH Science Bulletin piece co-authored with NIOSH Director John Howard.
• November 2025: This work was presented at the 2025 American Public Health Association conference (video available with APHA login).

Last updated August 14, 2026