The same idea as CSPM, aimed at the AI stack: continuously discover every model, dataset, pipeline, and agent running across your cloud and SaaS estate, then flag the misconfigurations, exposed keys, and shadow AI deployments before an attacker — or an auditor — finds them first.
Most platforms below combine these three capabilities — discovery is the common starting point, since you can't secure an AI asset you don't know exists.
Models, datasets, pipelines, and agents across cloud accounts, SaaS apps, and developer tools — including shadow AI nobody registered with security.
Exposed API keys, overly permissive access, unencrypted training data, and vulnerable model dependencies — scored and prioritized like any cloud posture finding.
Generate an AI bill of materials (AI-BOM), route findings into existing AppSec/CSPM workflows, and enforce guardrails before risky configurations reach production.
Eleven platforms spanning cloud-native AI-SPM, AI-aware data security posture, AI-BOM/application security posture management, and agent-specific posture discovery.
AI Posture connects context across infrastructure, identity, data, and AI to discover models and agents running across cloud environments and flag AI-specific risks like sensitive data exposure and exposed endpoints.
Agentless AI-SPM scanning across 50+ AI platforms and services — identifying misconfigurations, sensitive data exposure, shadow AI deployments, and exposed access keys without deploying sensors.
Combines data security posture management with AI agent security (Agent Guardian), covering data, identity, and action in one platform built to enable safe AI adoption.
AI-native application security posture management with an AI bill of materials (AI-BOM) and AI security testing (AIST), plus VibeGuard for securing code as AI coding assistants generate it.
Discovers, classifies, and governs every dataset an AI system can touch — from Copilot to Bedrock — at petabyte scale, closing the data-exposure gap that model-focused tools miss.
Inventories the mesh of AI coding assistants, agents, MCP servers, and plugins running across developer workstations, then enforces policy and detects prompt injection and data exfiltration in real time.
Extends its application security platform with an AI-BOM — a bill of materials for AI components — alongside behavioral testing and runtime protection for AI models and agents.
Falcon Cloud Security extends CrowdStrike's CNAPP to the AI stack — discovering models and pipelines across cloud environments and assessing them for misconfigurations and exposure.
Extends cloud-native application protection to generative AI applications, applying posture management across the AI application lifecycle from build to runtime.
Brings its runtime cloud security engine to AI workloads — detecting misconfigurations and anomalous behavior across AI pipelines running in containers and cloud infrastructure.
Builds a live inventory of AI agents and copilots, then evaluates each one's configuration and permissions against policy — validating which attack paths are actually exploitable and scoring the risk before agents go live.