BlockchainIntermediate#Compliance#Trust
KYA (Know Your Agent)
K
Definition
An informal checklist for inspecting an AI agent's identity, operator, scope, data, authority, and review path. KYA is not a regulated, certified, or universally accepted technical standard.
Source reviewed
- NIST: AI Agent Identity and Authorization opens in a new tab
Supports: NIST identifies agent and service identity, authorization, auditability, least privilege, and lifecycle management as key considerations for software and AI agents.
- NIST AI RMF 1.0 opens in a new tab
Supports: NIST's AI Risk Management Framework treats governance, transparency, accountability, security, and monitoring as context-specific risk-management work, not a generic badge of trustworthiness.
- OWASP: LLM06 Excessive Agency opens in a new tab
Supports: OWASP explains that broad authority and weak controls can let model output trigger damaging actions, supporting least privilege, approval, and action-boundary controls.
Concept Diagram
Key Takeaways
1
KYA is an informal due-diligence checklist, not a regulated standard or safety certification.
2
Identify the accountable operator, service identity, model or workflow version, tools, update owner, and incident path.
3
Disclose data sources, trust boundaries, requested actions, allowlists, value and rate limits, approvals, logging, pause, revocation, and recovery.
4
Keep financial, account, and irreversible actions behind independent policy checks and explicit human approval.
Practical Example
A research agent drafts a daily market brief. Its KYA record names the operator, read-only source APIs, input timestamps, model and workflow versions, prompt-injection tests, monthly cost cap, output archive, and incident contact. It cannot connect to a wallet, place an order, or change an account; a human reviews the cited brief before sharing it.
Finished learning?
Related Terms
AI
Agentic WalletA wallet or smart-account workflow that gives software an explicitly bounded ability to prepare or execute actions. It is an implementation pattern, not an ERC-4337 feature, safety guarantee, or reason to grant a model unrestricted signing authority.
AI
TEE (Trusted Execution Environment)A hardware-backed isolated execution environment that can protect specified code and data under an explicit platform threat model. Its assurance depends on the implementation, attestation policy, measured software, key release, configuration, and remaining hardware and software attack surface.
Blockchain
DAO (Decentralized Autonomous Organization)An organization or governance arrangement that uses blockchain-based rules, proposals, membership, and treasury controls to coordinate decisions or actions. Its voting, execution, transparency, and legal structure vary by design.
Blockchain
Smart ContractA program and its state deployed at a blockchain address, which runs its defined functions when transactions call it.
Related Deep Dives
Master the hub of Blockchain
Explore all our strategic guides about Blockchain to take your operations to the next level.
View all articles