Updated
Updated · OpenAI · Sep 16
OpenAI Finds Cross-Occupation AI Use Doubles to 25.9% as Workers Fold New Tasks Into Jobs
Updated
Updated · OpenAI · Sep 16

OpenAI Finds Cross-Occupation AI Use Doubles to 25.9% as Workers Fold New Tasks Into Jobs

3 articles · Updated · OpenAI · Sep 16

Summary

  • 25.9% of occupation-specific AI activity in July came from previously used cross-occupation tasks, up from 13.1% in April among about 6,200 workers tracked consistently.
  • More than 1.5 million work-related ChatGPT messages from April to July 2026 showed workers repeatedly returning to tasks outside their formal roles, suggesting AI use is moving from experimentation into routine workflows.
  • 23.6% of workers revisited a cross-occupation task the next month, versus 8.4% among comparable workers with no prior observed use, indicating prior AI-assisted task adoption strongly predicts recurrence.
  • 54% of workers returned to customer-discussion tasks the following month, 44% to advertising writing and 37% to marketing materials, while explaining financial information returned only 15% of the time.
  • OpenAI said the pattern points to job expansion before title changes, with work design—not just access to AI tools—likely shaping how companies deploy AI.

Insights

Are disappearing job boundaries a true productivity revolution, or simply a vendor-driven strategy to embed AI deeper into corporate workflows?
Could the ability to instantly borrow expertise through AI create a dangerous illusion of competence among non-specialist workers?
If AI handles the foundational tasks of junior employees, how will organizations successfully train their next generation of senior experts?