AI Summary

Microsoft has identified seven new failure modes in agentic AI systems, adding to those listed in its 2025 taxonomy. The expansion reflects four factors: rapid mainstream adoption of the technology, increased maturity of the Model Context Protocol ecosystem, the rise of computer‑use agents, and a growing body of empirical evidence from real‑world findings. These factors have revealed additional ways that agentic AI can malfunction or be exploited.

AI summaries can be wrong sometimes—always verify important details using the source article.