The increasing integration of agentic AI into business infrastructure poses a significant risk of insider threats, as these tools can be exploited by malicious actors or compromised due to inadequate security controls. Researchers at DTEX have identified a common workflow in Anthropic's Claude Cowork that can be leveraged to breach corporate environments1. This vulnerability highlights the evolving nature of attack methods, which may involve the manipulation of AI agents to gain unauthorized access to sensitive information. As AI tools become more pervasive, the potential for insider threats grows, underscoring the need for robust security measures to mitigate these risks. The potential breach of Anthropic's systems signals a shift in attack methods, which may have downstream regulatory and supply-chain implications. The ability of AI agents to be used as insider threats matters to cybersecurity practitioners, as it necessitates a reevaluation of existing security protocols to address this emerging risk.