THE Hacker News weekend feature argues that Model Context Protocols (MCPs) are enabling a new wave of AI agents that can access apps, data, and workflows across enterprises, turning AI into active, prompt-driven workers rather than mere chatbots. It describes identity dark matter as non-human identities that operate invisibly within hybrid environments, often outside traditional IAM controls, and warns they can pursue the path of least resistance with long‑lived tokens, API keys, and legacy credentials.
According to Gartner, the rapid enterprise adoption of Guardian/AI agents is outpacing governance and policy maturity, highlighting the governance gap. Team8’s 2025 CISO Village Survey found that nearly 70% of enterprises already run AI agents in production, with 23% planning deployments in 2026, and two‑thirds building them in-house.
The piece also stresses the need for tight integration of identity and information governance, advocate human sponsorship, time‑bounded entitlements, comprehensive auditing, and an enterprise‑scale supervisory layer to avoid unmanaged access becoming a systemic risk. It concludes that AI agents are here to stay, but it’s how they are governed that will determine security and trust.