Guides to the NIST publications that turn the AI RMF's secure and resilient characteristic into security work: the adversarial machine learning taxonomy, prompt injection, supply chain and agent guidance, SP 800-53 overlays, the Cyber AI Profile, secure AI development and NIST's open-source test platform.
What NIST AI 100-2e2025 covers and how security teams use it: predictive and generative AI attack classes, attacker objectives including misuse, the NISTAML identifiers, what is out of scope, and why NIST says mitigations remain empirical.
How NIST defines direct and indirect prompt injection, who is harmed in each case, the mitigation families AI 100-2e2025 describes and their limits, the design assumption it recommends, and the tools and benchmarks it names for testing.
What NIST says about attacks through training data, pre-trained models and plugins: the poisoning attack types, why AI supply chain risk needs more than software supply chain practice, the mitigations AI 100-2e2025 names and what the Generative AI Profile asks organizations to verify.
What NIST publications say about securing LLM-based agents as of October 2026: how AI 100-2e2025 defines agents and their added risks, prompt injection through tool output, the agent overlays planned in COSAiS and the design measures NIST describes.
What NIST's Control Overlays for Securing AI Systems project is building on SP 800-53, the five AI use cases it covers, the documents it draws on, and where drafts stood in 2026, for organizations that already run SP 800-53 or FedRAMP-style control programs.
What the NIST Cybersecurity Framework Profile for AI covers: its Secure, Defend and Thwart focus areas, its structure on CSF 2.0, draft status and 2026 working sessions, and how it relates to the AI RMF and the COSAiS overlays.
What SP 800-218A adds to the Secure Software Development Framework for teams that train, fine-tune or ship AI models: its audiences, how it is used with SP 800-218, its status after Executive Order 14110 was revoked, and where NIST builds on it.
What NIST's Dioptra platform is, how it is licensed and deployed, what it supports in the AI RMF Measure function, and what self-hosting it involves for teams that need reproducible AI security and robustness tests.
Tell us what you're working on, whether it's a deployment, an audit, a security test or a cyber range. You'll speak with an engineer who can help you scope it.