Phying News
P非影 中文
Curated security research, vulnerabilities, advisories and tools for practitioners.

How to Control AI Assets Before They Become Shadow AI

jfrog.com | blog | #ai-security | #ai-governance | #mcp | #jfrog | #coding-agents | #shadow-ai | #agent-guard

Summary

JFrog introduces Agent Guard, which governs the MCP servers, skills and plugins coding agents can reach through three enforcement layers built into the agent itself, each closing a different bypass.
Published
Collected

original ↗

Related coverage

blog · jfrog.com
The Evolution of JFrog AI Catalog: Your AI Control Plane for Agentic Development
JFrog AI Catalog evolves into an AI control plane for the Agentic Development Lifecycle: models, skills, plugins, and MCP servers become native artifacts, so agents consume only approved AI assets.
blog · jfrog.com
The Governance Gap: What IDC’s 2026 Data Reveals About AI and the Software Supply Chain
Drawing on IDC's 2026 survey of 1,000 organizations, JFrog panelists discuss AI as the top strategic mandate and why shadow AI has replaced shadow IT as a key unmanaged supply-chain risk.
blog · jfrog.com
Trusted AI Adoption (Part 2): Detection
Part two of JFrog's Trusted AI Adoption series argues continuous detection beats periodic scans, catching unregistered models, unvetted MCP servers and external calls agents pull 24/7.
blog · jfrog.com
Trusted AI Adoption (Part 1): Consolidation
First in JFrog's five-part Trusted AI series: a governed software supply chain coexists with an unmanaged AI shadow supply chain of models, MCP servers and scripts—pillar one is consolidation.
blog · jfrog.com
The AI Blind Spot Debt: The Hidden Cost Killing Your Innovation Strategy
JFrog frames 'AI blind spot debt' as the compounding cost of ungoverned AI: unvetted models, data leaking through personal API accounts, and MCP agents multiplying outside IT's view.
blog · jfrog.com
Beyond the Hype: Building a Future-Proof Foundation for the AI-Native Enterprise
JFrog ML's CTO on the AI-native enterprise: shadow AI, governance gaps and a 7x spike in malicious Hugging Face models show AI assets need the rigor applied to software supply chains.
back