Microsoft Defender email security benchmarking: Key insights from one year of data

Microsoft Security Blog posts (June 2026) cover multiple high-risk observations and guidance: email security benchmarking results for Microsoft Defender; ASSERT, an open-source framework to convert natural-language specs into executable AI evaluations; a telemetry-driven playbook for reconstructing AI activity in investigations; threat actors leveraging AI brands as social-engineering lures; a prompt-injection pathway in the Claude Code GitHub Action that could expose workflow secrets and the disclosure/mitigation timeline; an updated taxonomy of failure modes for agentic AI from a year of red