AI system cards for Red Hat products
AI models are only one part of what users actually interact with. Real-world value shows up when models are packaged into complete AI systems—with data pipelines, application logic, guardrails, monitoring, and user experiences.
Why AI system cards?
An AI system card is meant to help teams make informed choices by documenting how an AI system is built and how it’s operated—not just which model it uses.
- Architecture & components: what services, pipelines, and integrations make up the system.
- Models & data: which models are used and what data sources are involved (training, tuning, or augmentation).
- Security & safety posture: scope, guardrails, and how risks, like prompt injection, are addressed.
- Issues & change history: links to security and safety fixes and when they occurred (where applicable).
- Governance & licensing: contacts, accountability, and licensing (commonly CC BY 4.0 for the card text).
System cards
Ask Red Hat AI assistant
A conversational assistant in the Red Hat Customer Portal that helps users find relevant knowledge, docs, and support resources.
- Designed to speed up self-service troubleshooting and discovery of approved Red Hat content.
- This system card documents purpose, scope, core components, data sources, security, and safety considerations.
Red Hat OpenShift Lightspeed
An AI-powered assistant integrated into the OpenShift web console to help users learn, troubleshoot, and investigate cluster resources.
- Uses official OpenShift documentation for augmentation and provides guidance via natural language prompts.
- This system card covers intent and scope, platform details, provider options, and security considerations, such as prompt injection defenses.
Red Hat Enterprise Linux Lightspeed
AI-assisted guidance for common RHEL workflows, for example, plain-language help in the command line, to simplify operations for new and experienced admins.
- This system card covers intent and scope, platform details, data provenance as well as some safety and security considerations for the product