Solutions  ·  2026-07-22

Box unveils AI agent security and governance controls (guardrails, prompt-injection detection, MCP guardrails)

SolutionsHigh impactGlobal
Box announced (2026-07-21) new enterprise controls covering both Box-native and third-party agents (Claude, ChatGPT, Gemini): agent guardrails tied to content sensitivity, prompt-injection detection screening inputs before they reach a model, MCP-server-specific permission scoping, classification-based access walls, activity oversight/audit trails, and human-in-the-loop approval for high-impact actions.
Box cites 90% of IT leaders naming security/trust as the top blocker to agentic AI adoption on their content; this embeds AI-SPM-style controls directly at the content layer for a platform with a massive enterprise content footprint, addressing MCP and cross-vendor agent risk in one release.
Enterprises using Box (financial services, healthcare, legal, insurance) planning agentic workflows against Box content should plan onboarding as it rolls out to Enterprise Advanced customers over the coming months.
SiliconANGLEBusiness Wire / Box
See this in the live feed Explore related AI security and governance findings — updated every morning.
Open the feed →