A recent analysis of over 800,000 messages from U.S. ChatGPT users reveals a fundamental shift in how artificial intelligence is restructuring workplace roles. Titled Work at the Frontier: How AI is Expanding What People Do at Work, the report introduces the concept of task crossover, describing a pattern where employees increasingly perform duties historically confined to other occupations. While 16.8 percent of all work-related AI interactions cover general professional activities, 43.5 percent of occupation-specific prompts involve tasks that fall outside the user’s primary role. This phenomenon indicates that AI is functioning as a bridging tool, allowing professionals to bypass traditional departmental handoffs. Small-business owners are independently drafting marketing copy and conducting financial analysis. Sales representatives are exploring customer datasets typically reserved for analysts. Marketers are resolving technical website issues without developer support. Consequently, the division of labor is being reorganized in real time, with AI enabling workers to experiment with new task combinations before formal job descriptions or corporate structures adapt. The research, part of a new Work at the Frontier series and grounded in the AI Jobs Transition Framework, highlights that crossover is not evenly distributed. Marketing and engineering tasks exhibit the widest reach, frequently appearing in prompts from professionals across design, finance, human resources, legal, sales, and customer experience. Designers absorb the highest volume of external tasks, with 35.2 percent of their usage tied to other fields, though their own output rarely migrates outward. Engineering operates in reverse, supplying technical troubleshooting and system management skills to other departments while absorbing fewer external duties. Marketing stands out for bidirectional exchange, with over a quarter of its users engaging in outside tasks and its own outputs reaching other roles at the highest rate in the dataset. Company structure significantly influences crossover patterns. Workers in smaller organizations, where specialized teams are often unavailable, rely more heavily on AI to cover skill gaps. Among average users, external task usage drops from 18.9 percent in workspaces of two to five seats to 16.3 percent in firms with over a hundred employees. This suggests AI acts as a critical generalist resource in lean environments. Heavy users, however, show less variation by company size, indicating they have established stable, role-specific AI workflows that transcend organizational boundaries. These usage metrics provide an early, granular signal of occupational evolution. By revealing how AI redistributes task ownership across professions, the data allows researchers and policymakers to anticipate workforce restructuring before it manifests in traditional labor market statistics. As AI continues to lower the barriers to specialized skills, the traditional boundaries of occupation will likely continue to blur, fundamentally altering how enterprises allocate labor and define professional expertise.