Claude Tag, an AI system for Slack, aims to make company knowledge searchable but creates a double bind for employees who are simultaneously warned of AI-driven job displacement. This dynamic incentivizes employees to withhold candid information, leading the AI to capture only a "sanitized" version of company knowledge. The author questions whether companies should sacrifice candor and control over their "real IP" for AI productivity.
Enterprise AI agent adoption faces a human-incentive problem, as employees may withhold critical knowledge from systems that threaten their jobs.