Requirement-centered standards map
Start with an enterprise AI security requirement, then see which standards, safeguards, control objectives, and threat references can help you evaluate it.
Vendor-focused view
0 security requirements with strong or limited public support determine which rows appear. This does not mean the vendor implements or conforms to a framework.
How to read this map
References share a row because they help address the same requirement. They are not interchangeable. Each label explains how closely a reference aligns with that requirement.
Standards and framework connections
Important references remain visible without hover. Open a framework research view for market-wide context, or start from a vendor page to see only references connected to that vendor's public claims.
CIS Controls v8.1
The research team selected the CIS safeguard connections shown here; they are guidance, not official CIS mappings. The official 2026 companion guides explain how CIS Controls apply to three AI environments.
Applies CIS Controls v8.1 to prompts, context handling, model access, sensitive data, and LLM operations.
Open official guide ↗AgentApplies CIS Controls v8.1 to agent planning, tool use, delegated actions, identity, and runtime behavior.
Open official guide ↗MCPApplies CIS Controls v8.1 to MCP tool access, non-human identity, authorization, logging, and protocol operations.
Open official guide ↗Source guide
Use each source for the job it is designed to do. Vendor implementation and product effectiveness still require separate evidence.
AI RMF 1.0 Playbook
AI risk management reference for governance, mapping, measurement, and lifecycle management. Useful for structuring diligence questions, but not a standalone selection model, certification proxy, or substitute for enterprise-specific risk, architecture, and operating requirements.
v2.0 (NIST CSWP 29, February 2024)
Cybersecurity outcome lens (GOVERN, IDENTIFY, PROTECT, DETECT, RESPOND, RECOVER). NIST IR 8596 Cyber AI Profile deferred until it exits draft; subcategory IDs in this project map to CSF 2.0 directly and remain stable when the AI Profile lands.
v8.1 with 2026 AI Companion Guides
Prioritized cyber-hygiene safeguards with AI-specific interpretation for large language model (LLM), agent, and Model Context Protocol (MCP) environments. This project maps normalized AI security requirements to relevant CIS safeguards as a research interpretation; it does not present an official CIS mapping, implementation assessment, or certification result.
ISO/IEC 42001:2023
AI management system lens for procurement diligence prompts and auditor discussion. Mappings are indicative clause/control alignments only; they are not a certification proxy, vendor assurance substitute, or evidence that a product satisfies ISO/IEC 42001.
Q2/Q3 2026
Market and lifecycle lens for large language model (LLM) and generative AI AppSec/SecOps capabilities.
Q2/Q3 2026
Market and lifecycle lens for agentic AI, agent-to-agent (A2A), tool, memory, non-human identity (NHI), and runtime controls.
2026.06
Adversary-behavior lens for AI systems. Technique and mitigation references describe threat paths and defensive hypotheses; they do not establish that a vendor prevents, detects, or responds to the behavior.
2025 (v2.0)
Application failure-mode lens for large language model (LLM) systems. Mappings identify risks a requirement should address and are not proof of product efficacy or OWASP endorsement.
2026
Agentic failure-mode lens covering autonomy, tools, identity, memory, multi-agent communication, cascading effects, and oversight.
v1.1 (June 22, 2026)
Vendor-neutral control-objective lens. This project uses the official AI Customer implementation publication with 247 controls across 18 domains; a mapped control is a diligence and test target, not evidence of implementation.
O-RA 2.0.1 / O-RT 3.0.1
Business-loss scenario and quantitative risk-analysis structure. Scenario templates are modeled, but event frequency, vulnerability, and loss magnitude remain explicit enterprise inputs.
Point-in-time public research
Licensing and packaging context. Not a security framework.
This map organizes research. Certification, audit conclusions, framework conformance, vendor control implementation, and product effectiveness require separate evidence.