A RedAccess study finds 380,000 publicly accessible assets built with vibe coding tools, with around 5,000 containing sensitive corporate data. Verified examples include unredacted customer chats, medical trial listings, incident response records, and exposed bank information. The issue stems from platform defaults that publish apps unless manually secured, turning shadow AI into a production risk layer.
Microsoft has pulled Agent 365 out of preview and into general availability, betting that autonomous “shadow AI” is already a day-to-day enterprise risk. The platform offers a unified way to inventory and govern AI agents across clouds, SaaS, and endpoints—plus discovery of local agents on managed Windows devices, with Defender-style “blast radius” context.
Your news, in seconds
Get the Beige app — every story in 60 words, updated hourly. Free on iOS & Android.
Swipe through stories, personalise your feed, and save articles for later — all on the app.
