OSHA's New Predictive Safety Framework Could Make AI Monitoring Mandatory by 2028

The Occupational Safety and Health Administration published a 47-page draft framework this week outlining how predictive AI systems could become a required component of workplace safety programs in high-risk industrial environments. If finalized, the rule would take effect in 2028.

The framework, titled "Predictive Safety Systems in Automated Industrial Environments," defines three tiers of AI monitoring requirements based on facility risk classification. Tier 1 facilities — including chemical processing, heavy manufacturing, and energy generation — would need real-time AI systems capable of detecting unsafe conditions and triggering automated interventions.

Industry response has been divided. The National Association of Manufacturers called the timeline "aggressive but achievable," while the American Petroleum Institute warned that compliance costs could exceed $2 million per facility for Tier 1 operators.

Several vendors are already positioning for the mandate. Honeywell's Safety Watch platform, Siemens' Industrial Safety Suite, and startup Voxel AI all offer computer vision systems that monitor worker behavior and environmental conditions in real time.

The public comment period runs through August 2026. A final rule is expected by Q1 2027.