TrojAI and Wiz: Closing the Loop on AI Security
TrojAI is now integrated with Wiz’s AI-APP, connecting Wiz’s AI asset discovery with TrojAI’s AI security testing and runtime protection
Key Takeaways
- AI red teaming has moved from testing single models to simulating attacks across the full stack, because risk emerges where components connect
- Agentic AI has widened the attack surface, turning small prompt manipulations into real-world consequences across chained actions
- Adversarial testing is now an operational discipline across government, labs, and enterprises, with continuous simulation replacing one-off tests
- Traditional security falls short because AI systems are probabilistic, context-sensitive, and always evolving
- The path forward is continuous, system-level security built into the AI lifecycle, not a one-time audit
TrojAI is pleased to announce our Wiz integration as part of Wiz’s AI Application Protection Platform (AI-APP) launch. At a high level, the integration connects Wiz’s AI asset discovery with TrojAI’s AI security testing and runtime protection, bringing together a continuous AI security workflow around visibility, testing, and protection.
What is the TrojAI + Wiz Integration?
As enterprises continue to deploy AI apps and agents, they are increasingly challenged with both visibility and security of their AI estate. TrojAI and Wiz address these concerns through a closed-loop system centered around discovery, testing, and protection.
Wiz identifies and inventories AI models and endpoints across your environment, while TrojAI ingests that inventory to perform automated red teaming. Once inventoried, TrojAI Detect’s agent-led red teaming leverages adversarial agents to test models for risks like prompt injection, data leakage, and unsafe behavior. The results are then pushed back into Wiz, enriching its security graph with AI-specific findings and context. With better context on AI risks, TrojAI users can actively monitor and protect their AI in runtime with TrojAI Defend.
Why it Matters
Most enterprises today face a fragmented AI security landscape. They lack a clear inventory of AI systems. They struggle to test models for real-world vulnerabilities and want to accelerate their AI deployments. They don’t have a unified way to prioritize and remediate risks.
This integration addresses these concerns. By combining Wiz and TrojAI, security teams gain full visibility into AI assets and can continuously identify exploitable risks like jailbreaks or data exfiltration. From here, they can use the Wiz console to prioritize threats based on real attack paths and cloud context, and leverage TrojAI Defend to monitor and protect their AI assets with runtime policies that actively block against AI threats like prompt injection, data leakage, and toxic content.
How it Works
- Discover: Wiz scans the environment to identify AI models and endpoints.
- Test: TrojAI pulls this inventory and runs automated red teaming to uncover vulnerabilities like prompt injection or sensitive data exposure.
- Enrich and prioritize: Findings are pushed back into Wiz, where they are correlated with cloud infrastructure context to identify real attack paths.
- Protect: TrojAI enforces runtime guardrails to prevent potential exploits, blocking malicious prompts, redacting sensitive data, and monitoring live interactions.
The Bigger Picture
Wiz tells you where your AI is and how it connects to your cloud risk. TrojAI tells you how that AI can be attacked and gives you the tools to stop it in real time. By combining Wiz’s cloud and AI visibility and TrojAI’s deep AI security testing and enforcement, organizations can finally move from reactive security to a proactive, lifecycle-based approach, securing AI from development through production.
FAQs
The TrojAI and Wiz integration is a closed-loop AI security workflow that connects Wiz’s AI asset discovery with TrojAI’s automated red teaming and runtime protection. Wiz identifies and inventories AI models and endpoints across the environment, TrojAI ingests that inventory to test for vulnerabilities, and the findings are pushed back into Wiz to enrich its security graph with AI-specific risk context.
The integration addresses the three core gaps enterprises face: lack of AI asset visibility, inability to test models for real-world vulnerabilities, and no unified way to prioritize and remediate risks. By combining Wiz’s cloud and AI visibility with TrojAI’s security testing and runtime enforcement, security teams gain continuous coverage from discovery through protection, eliminating the blind spots that fragmented tools leave behind.
Wiz scans the enterprise environment to identify and inventory AI models and endpoints, providing a comprehensive map of the AI estate. This inventory is then shared with TrojAI, enabling automated security testing against every discovered asset. Findings from TrojAI are pushed back into Wiz, where they are correlated with cloud infrastructure context to help teams identify real attack paths and prioritize remediation.
TrojAI pulls the AI asset inventory from Wiz and runs automated, agent-led red teaming against each discovered model or endpoint — testing for risks like prompt injection, data leakage, jailbreaks, and unsafe behavior. The results are then pushed back into Wiz’s security graph, enriching it with AI-specific findings that can be correlated with broader cloud risk context for accurate prioritization.
The integration is available through the Wiz Integrations Marketplace. Visit the TrojAI integration page on Wiz to get started. For guided onboarding or to learn more about how the integration fits your environment, you can also book a demo with the TrojAI team.
