On September 16 (local time), OpenAI's economic research team unveiled its second Frontier Work report. This report delved into an analysis of over 1.5 million work-related ChatGPT messages exchanged by registered users in the United States from April to July 2026. The findings revealed a steady increase in the proportion of employees utilizing AI to tackle tasks that fall outside their primary job descriptions. Among the roughly 6,200 employees engaged in this ongoing study, the proportion of cross-occupational tasks within AI workloads nearly doubled, surging from 13.1% in April to 25.9% in July. The report highlighted that when employees are confronted with cross-occupational tasks, they tend to furnish AI with contextual information and perform verification checks, rather than seeking explanations, operational guidance, or specific formats from the AI. This indicates that employees might be harnessing AI's professional prowess to accomplish tasks that are unfamiliar to them. Furthermore, notable disparities were observed in the sustained usage levels across various tasks. Tasks involving customer communication, advertising, and promotional copywriting exhibited return rates as high as 54% and 44%, respectively. In contrast, the return rate for legal research tasks stood at a mere 10%.