Agent正在重写分工，而治理框架还停在石器时代

As labor roles evolve into role-based units, the governing framework remains in a Stone Age. This transformation is evident from three key indicators. Firstly, the European Union's AI law (Regulation (EU) 2024/1689) took effect on August 1, 2024, defining high-risk AI systems that encompass hiring, credit, and public infrastructure, imposing stringent requirements on risk management, data governance, logging, and human supervision.

Secondly, enterprise software is elevating agents to employee status. Salesforce launched Agentforce in September 2024 and is aggressively promoting it as a "digital workforce" product line in 2025. This platform allows businesses to define agents for specific roles using natural language and monitor their behavior across business systems.

Thirdly, the structural warning for the labor market has been publicly unveiled. A recent World Economic Forum report on the future of employment in 2025 surveyed business decision-makers from major economies, shifting the focus from "the number of jobs AI will replace" to "the redefinition of tasks and people alignment." This task redefinition is precisely where agentized division of labor operates: by not replacing people but by replacing many decision-making steps within a job with automated agents.

These three anchor points intersect to highlight the core issue of responsible agent economics. However, this change has not waited for full regulatory completion before occurring. In the past 18 months, numerous organizations have begun to introduce agents in a haphazard manner, such as handling customer service tickets, generating automatic email drafts, and granting agents direct read access to internal databases.

While technical teams focus on ensuring pipelines run smoothly and business teams seek KPI improvements, legal and compliance teams remain unclear on where to establish human oversight points, creating a gap that could lead to loss of control.