What AI Red Teaming (NIST AI 100-4) requires
Adversarial red teaming constitutes a mandatory control for designated AI systems, aligning with directives in U.S. Executive Order 14110 and fulfilling the accuracy, robustness, and cybersecurity requirements detailed within the EU AI Act's Article 15. This node’s primary objective is to systematically identify, classify, and mitigate vulnerabilities through structured testing cycles conducted every 90 days, an operational tempo that supports the MEASURE function of NIST's AI Risk Management Framework. Each cycle must employ a minimum of 5000 adversarial prompts designed to stress-test system defenses against a comprehensive range of threats articulated in the NIST AI 100-4 taxonomy. The protocol mandates active simulation of evasion attacks, data poisoning scenarios, and model extraction attempts. Performance is evaluated against stringent thresholds, requiring a jailbreak success rate not to exceed 0.05 and a minimum robustness confidence score of 0.9. Testing specifically targets critical OWASP Top 10 for LLM vulnerabilities, including LLM01 Prompt Injection and LLM06 Sensitive Information Disclosure. To ensure procedural integrity consistent with ISO/IEC 23894 guidance, all evaluations require human-in-the-loop testing conducted by an operationally independent red team. Upon discovery of a critical vulnerability, an automatic quarantine protocol is triggered to prevent further exposure or compromise.
Pillar: AI Governance & Law · Authority: NIST (National Institute of Standards and Technology) · Version: 1.1.0 · Last updated:
Primary source: https://doi.org/10.6028/NIST.AI.100-4
SHA-256 integrity: 824bb6cc9a3047fe303e4605c249730560d116b34f8b6bae3fa9ce5c05166c5c
Primary Citations — 6 traced to source
- NIST AI 100-4: Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations
- U.S. Executive Order 14110: Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (Section 4.1: AI Red Teaming)
+ 4 more citations (full bibliography, deterministic workflow, actionable schema and crosswalks) included in the vault unlock — $0.01 via Skyfire / L402 / Direct Base USDC.
Access
- Discovery (free): /api/v1/nodes/nist-ai-100-4-redteam.json — 6-field metadata
- Vault (full node): /api/v1/vault/nodes/nist-ai-100-4-redteam.json — full 13-key payload, $0.01 USDC (L402/Skyfire/Direct Base)
- Canonical URL: https://bidda.com/intelligence/nist-ai-100-4-redteam
- Back to registry: Browse all 10,090 compliance nodes