Glossary
CWPP Report Glossary¶
Workloads Protected: Refers to the no. of workloads where the policies are applied
Too many policy violations: Count of workload which has more than 100 unique policy violation
List of workload with no protection: Shows list of workloads where there is no policy applied
List of workload with policy violation: Shows list of workload where policy violation has occurred and alerts have been generated for applied policies
AI Security Glossary¶
Definitions for AccuKnox's eight AI security modules and their related product names.
AI Security Posture Management (AI-SPM): Continuous discovery, inventory, and posture scoring of AI/ML assets (models, datasets, notebooks, agents) across cloud and on-prem, including shadow-AI detection.
AI Detect & Respond (AI-DR): Detection and response for AI security incidents, unauthorized AI deployments, and runtime threats correlated across cloud control planes.
Agentic AI Security: Runtime sandboxing, tool-call enforcement, and least-privilege controls for AI agents and agent toolchains, delivered through KubeArmor runtime enforcement.
AI Model & Dataset Security: Static and supply-chain scanning of models and datasets (Hugging Face, GitHub, registries) for malware, tampering, and unsafe serialization such as pickle code injection.
AI Red Teaming & Pen Testing: Automated adversarial testing of LLMs and ML models for prompt injection, jailbreaks, toxicity, hallucination, and data-leak vulnerabilities.
AI Guardrails (Prompt Firewall): Real-time inline inspection and enforcement on LLM prompts and responses to block injection, jailbreaks, sensitive-data leakage, and unsafe output.
AI Identity Security: Identity and entitlement controls for AI agents, models, and non-human identities. Coming soon.
AI Compliance & Governance (AI-GRC): Governance, risk, and compliance mapping for AI systems against frameworks such as the EU AI Act, NIST AI RMF, and ISO 42001. Coming soon.
Prompt Firewall: The product that delivers AI Guardrails.