←Back to NewsAI News/AgentsnewsAgentsSecurityMicrosoft's run-assert-eval Cuts AI Agent Cross-Account Failures From 43.8% to ZeroMicrosoft's new run-assert-eval skill turns agent risk discovery, evaluation, policy generation, and re-testing into a single VS Code prompt.SourceMicrosoft DeveloperPublishedSep 24, 2026, 5:37 PMAuthorAlphaSignal NewsroomRead1 min readMicrosoft's new run-assert-eval skill turns agent risk discovery, evaluation, policy generation, and re-testing into a single VS Code prompt.Reporting is indexed from AlphaSignal. Rights remain with the original publisher and cited sources.Read original report ↗Next readsAlphaSignal · paperAI Agents Secretly Upsell Wealthy Users, Turning a $91 Flight Into $601Google DeepMind · paperGoogle DeepMind's Gemini Agents Spontaneously Split Into Cheaters and WhistleblowersAI at Meta · newsMeta's Muse Agent Books Travel and Negotiates Bills With Your Accounts
