AI Security firms

1. Security for AI (Protecting Models and AI Agents)

These companies focus specifically on testing models for vulnerabilities, stopping prompt injections, detecting data poisoning, and governing "Shadow AI". 

Protect AI: Protect AI provides a comprehensive platform that secures AI/ML applications from model selection and testing to runtime.
HiddenLayer: HiddenLayer offers runtime AI security to protect against adversarial attacks, model manipulation, and data leakage.
Mindgard: Mindgard specializes in continuous security testing and autonomous red-teaming for AI systems.
Zenity: Zenity is an AI agent-centric security and governance platform that monitors and restricts what AI agents are allowed to access and do.  

2. AI-Powered Cybersecurity (Securing the Enterprise)

These platforms utilize machine learning and AI for threat detection, behavioral analysis, and autonomous incident response.

Wiz: Wiz acts as an AI and cloud security posture manager that identifies exposed AI infrastructure, data, and runtime threats.  
SentinelOne: SentinelOne features a massive-scale AI data lake and "Singularity AI SIEM" for real-time threat detection and response.  
CrowdStrike: CrowdStrike utilizes AI for endpoint protection, but also features modules dedicated to securing enterprise AI systems and discovering Shadow AI.
Vectra AI: Vectra AI uses behavior-based AI algorithms to detect hidden attackers and anomalous movements across hybrid cloud and identity networks. 

 AI Container Security

AI security for containers protects the application software, models, and environments from supply chain vulnerabilities, malicious workloads, and adversarial AI attacks. Leading enterprise platforms combine cloud-native protection with specialized AI capabilities. 

Leading AI Security & Container Platforms

Aqua Security: Offers dedicated Aqua Secure AI to protect AI workloads directly inside the containerized environments where they process data. It provides deep visibility, runtime protection, and enforces policies mapping to the OWASP Top 10 for LLMs. 
Palo Alto Networks (Prisma Cloud): Secures AI and container runtimes across their full lifecycle. Its platform offers AI Runtime Governance to govern AI agent chains, detect threats, and apply defense-in-depth security to microservices architectures. 
Orca Security: Provides agentless Cloud-Native Application Protection Platform (CNAPP) capabilities, including advanced reasoning for AI agents and dynamic risk correlation from code to cloud. 
SentinelOne: Integrates Singularity Cloud Workload Security with AI-powered threat detection to defend container workloads and enforce compliance standards in real-time.
Sysdig: Utilizes eBPF-based runtime visibility Container and Kubernetes Security to detect unusual container behavior and stop threats. It includes AI-guided remediation to fix vulnerabilities efficiently with minimal disruption. 

Key Open-Source & Specialized Solutions

DockSec: An open-source, AI-powered tool combining container security scanners (like Trivy) with language-model layers to provide line-specific fixes for Dockerfiles.
HiddenLayer: Delivers a HiddenLayer AI Security Platform that monitors, detects, and responds to adversarial threats on generative and agentic AI models in real time.
Chainguard: Provides hardened, minimal Chainguard container images to secure the AI software supply chain, eliminating engineering hours typically spent dealing with vulnerabilities.