red teaming

Microsoft’s AI Red Team updated its agentic AI failure-mode taxonomy on June 4, 2026, adding seven categories after a year of red-team engagements against deployed agent systems, with new emphasis on supply-chain compromise, tool abuse, visual attacks, session contamination, and human-approval...

Microsoft is urging security teams to red-team AI systems across the entire application stack, not just the model, with Microsoft red teaming executive Craig Nelson emphasizing data connections, backend automation, credentials, and logging in a recent Microsoft Inside Track security video. The...