Agentic AI, Red Hat OpenShift, and NVIDIA: Shifting to precision security

Red Hat is pioneering the use of agentic AI to shift vulnerability management from volume to precision. By combining the security-hardened foundation of Red Hat OpenShift with advanced AI frameworks from NVIDIA, we're delivering actionable security intelligence that provides genuine business value for the enterprise. 

Here's the problem: A single software package can trigger hundreds of alerts because one imported function or library within it is theoretically vulnerable. Traditional scanners stop at the package level, generating a flood of false positives for vulnerabilities that are never actually reachable in your application. Teams are forced to waste countless hours triaging risks that don't exist in practice.

This is where Red Hat Trusted Profile Analyzer, which includes the exploit intelligence feature, fundamentally changes this dynamic.

It doesn't stop at the package. Exploit intelligence's agentic AI performs a deep, function-level analysis of your actual application code paths. It traces the execution flow to definitively determine: Is the specific vulnerable function reachable and executable in your application's context?

If the function is not reachable, not executable or otherwise mitigated, exploit intelligence automatically declares the vulnerability  is not exploitable in your environment. This shift from package-level noise to function-level exploitability analysis is a genuine breakthrough. The result is simple and transformative: your security teams stop chasing hundreds of theoretical vulnerabilities and can focus their efforts exclusively on those that are truly exploitable and pose a real business risk.

Delivering a  premier platform for AI and traditional apps

As organizations scramble to adopt generative AI, they face a critical challenge in building and deploying these models securely without data leaving their environments. Red Hat OpenShift AI serves as the application development platform for both traditional workloads and AI-infused applications, so teams can more securely deploy AI in line with digital sovereignty requirements.

Through our collaboration with NVIDIA, Red Hat is bringing enterprise AI to life. This collaboration has rapidly evolved from hardware operators to co-engineered solutions, including the integration of NVIDIA NIM into OpenShift AI, DGX hardware validation, and joint offerings like the Red Hat AI Factory with NVIDIA. By taking these AI blueprints and running them on OpenShift, we empower platform engineers to harness the power of AI more safely behind their own firewalls.

Exploit intelligence: Business value through agentic precision

The clearest example of this agentic AI value is exploit intelligence, a capability built on the NVIDIA Morpheus Agent and NVIDIA NeMo Agent Toolkit.

Inside the exploit intelligence agentic workflow

Exploit intelligence implements a sophisticated "plan-and-execute" agentic workflow to deliver human-verifiable proof. Here is how the AI acts on behalf of the security analyst:

1. Input and context ingestion: Exploit intelligence accepts a container image’s software bill of materials (SBOM) and a target common vulnerability and exposure (CVE). It then pulls the corresponding CVE intelligence from various authoritative sources and incorporates user-specific details if present. It then pulls the container’s contents and prepares for the inspection process.
2. Large language model (LLM) planning: Based on the collected vulnerability intelligence gathered from public sources (National Vulnerability Database, GitHub, Red Hat’s security data streams, and more), the planning agent generates a context-sensitive task checklist. This checklist defines the exact steps needed to verify if the CVE is exploitable in this specific application.
3. Agent execution: An agent iterates through the checklist. Red Hat has developed code analysis tools to find function patterns and explore relationships in the codebase, gathering precise evidence regarding function reachability, execution paths, environmental configurations, and other controls in place.
4. Synthesis and verdict: A classification and reasoning agent condenses the findings and arrives at a verdict. For example, if the code is not reachable, it assigns a "Non-Exploitable" status and generates a clear, human-readable justification (code_not_reachable or protected_by_compiler, depending on code execution path and vulnerable function args).

Integration with the trusted software factory 

Agentic AI delivers the most value when deeply integrated into the developer ecosystem. Exploit intelligence is a critical feature of Red Hat Advanced Developer Suite and works alongside the trusted software factory.

• Red Hat Trusted Profile Analyzer: Trusted Profile Analyzer manages the risk profile of your software via SBOMs. When Trusted Profile Analyzer detects a CVE, it can trigger exploit intelligence. If exploit intelligence determines the CVE is a false positive, it automatically generates an industry-standard CSAF VEX (Common Security Advisory Framework Vulnerability Exploitability eXchange) file. Trusted Profile Analyzer ingests this VEX file to silence unnecessary churn.
• Red Hat Trusted Artifact Signer: While exploit intelligence analyzes code context, Trusted Artifact Signer cryptographically signs software artifacts and AI models, helping provide provenance and assurance that the code analyzed by exploit intelligence hasn't been tampered with before deployment.
• Trusted software factory: A trusted software factory reference implementation that mirrors Red Hat’s internal build system brings the same rigorous security standards Red Hat uses for its own products to enterprise customers via Red Hat Advanced Developer Suite.
• Red Hat Advanced Cluster Security for Kubernetes: The value of exploit intelligence extends directly to your runtime environment through Red Hat Advanced Cluster Security integration. Red Hat Advanced Cluster Security can mark CVEs as "false positives" so no alerts (violations) are generated. This can reduce the "noise" and let security teams focus on those CVEs which pose real risks. 

By leveraging Red Hat OpenShift, NVIDIA's AI frameworks, and the comprehensive Rat Hat Advanced Developer Suite portfolio, platform engineers can cut through the noise. Exploit intelligence proves that agentic AI can be a practical and highly efficient tool that drastically improves developer velocity and fortifies the enterprise software supply chain.

Learn more: Red Hat Advanced Developer Suite.