何が起きたか
On Nov 10, 2026 Orca Security announced expanded capabilities: AI AppGen Security now detects digital fingerprints left by AI app builders (Lovable, v0, Replit, Base44, Bolt) plus hosting services (Vercel, Netlify, Supabase) to discover shadow AI apps agentlessly via SideScanning, correlates them with IAM/identity/access to show data exposure, and adds AppGen support for Base44, OpenAI and Vercel with posture checks. It also adds AI agent security posture: an Agents tab in the AIBOM to inventory cloud-managed and self-hosted agents (Claude Code, Codex, Gemini CLI), map them to models/MCP servers/identities/data, and flag over-privileged agents, disabled approval controls, exposed MCP credentials, and insecure HTTP.
なぜ重要か
Addresses the breakout 'shadow AI app' and agent-access blind spots where non-engineers build with AI platforms and agents act on over-privileged identities — giving security teams discovery and blast-radius context without extra agents.
適用範囲
Cloud and AI security teams should evaluate Orca's AI AppGen and agent posture capabilities for discovery/risk-prioritization of citizen-built AI apps and coding agents, and use the AIBOM agents view for agent inventory and access mapping.