AI & Safety · Research brief
AI and worker safety: potential, limits and questions
A source-linked discussion of where AI may affect occupational safety, and how to assess its limitations before relying on it.
Two questions belong in the same conversation
AI for safety: could a tool help organise information, detect a pattern or support a decision? Safety of AI: what new risks could the tool introduce through errors, missed events, unreliable data, privacy impacts, cyber incidents or changed work practices?
NIOSH researchers have written that the adoption of AI technologies can result in a broad range of hazards and risks to workers, including those linked to industrial robotics and algorithmic management [1]. They list workplace ethical principles. Autonomy, privacy, freedom from bias, and transparency with accountability, and propose options such as upskilling practitioners, joint developer-practitioner evaluations, independent audits, certification and structured safety evidence such as safety cases [1].
An output is not a control
An alert or prediction does not demonstrate that a hazard has been controlled. Performance can vary with the people, equipment, lighting, environment and operating conditions represented, or missing. In the data used to build and test a system.
Questions before a workplace pilot
- What specific decision or task is the AI intended to support, and what evidence shows it can do that in this context?
- What are the consequences of a missed detection, false alarm, outage or incorrect recommendation?
- Who verifies outputs, has authority to stop work, and maintains non-AI fallback arrangements?
- How are workers involved, and what privacy, retention, access and cybersecurity controls apply?
- What measures will show whether the system improves outcomes without creating new exposures or burdens?
Risk-management frameworks
The NIST AI Risk Management Framework organises risk work under four functions. Govern, Map, Measure and Manage [2]. NIST describes the framework as voluntary; it is not a safety certification or proof that a particular application is safe [3].
General educational material, not legal advice, a site-specific risk assessment or a work authorisation. Confirm current requirements in your jurisdiction and follow site procedures and competent advice.
References
- NIOSH Science Blog. Exploring Approaches to Keep an AI-Enabled Workplace Safe for Workers (9 Sept 2024)United States
- NIST AIRC. AI RMF CoreUnited States
- NIST. Artificial Intelligence Risk Management Framework (AI RMF 1.0)United States
All sources opened and checked on 30 October 2026. Requirements and guidance change; always consult the current version.
Editorial status: source-checked research brief. No independent external technical review is claimed. Spotted an error? See corrections.
