Secure an AI agent as an identifiable, authorized workload—not as an employee account it happens to use. Give each agent a distinct identity tied to a responsible person or system, limit its authority to the task, protect and expire its credentials, and monitor its actions across tools and data. Existing identity and cybersecurity controls provide a practical foundation, but NIST’s current work does not establish a mature, universally settled standard for agent identity and authorization.
Why AI agents change the identity problem
A conventional application usually acts through a defined service identity. An agent can plan, call tools, retrieve data and take actions in sequence, sometimes responding to information it encounters along the way. NIST’s Center for AI Standards and Innovation (CAISI) described agent systems as capable of “planning and taking autonomous actions that impact real-world systems or environments” in its January 12, 2026 announcement.
That creates familiar security risks at a different pace and scale. If an agent uses an employee’s login, it may inherit more access than its task requires, blur who is accountable for an action and make identity governance harder. If it encounters malicious instructions in a document or other input, it may try to use its permitted tools in an unintended way. Identity controls cannot determine whether every model decision is safe, but they can define what the agent is allowed to access and do.
Build agent security on five controls
1. Give each agent an accountable identity
Assign each agent or agent workload a distinct identity rather than sharing an employee’s credentials. Bind that identity to the human or system responsible for operating it, and record its owner, intended function, environment and lifecycle status. This improves attribution when an agent acts, delegates work or is retired. It also makes it possible to review and revoke access without confusing the agent’s activity with the operator’s own login.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →#1 Best Overall
- A FIDO security key with PUF technology provides a unique, hardware-rooted trust anchor that resists tampering and cyber attacks, offering stronger security than conventional designs.
- FIDO2 Certified Protection – Enjoy phishing-resistant security with FIDO2 certification, ensuring top-tier account safety across Windows, macOS, Linux, iOS iOS, Android and more.
- Easy to use & Portable – Designed with a compact USB-C interface, Clife key fits easily on your keychain for secure access anywhere. Simply plug in and authenticate with ease.
- Universal Compatibility – Works seamlessly with hundreds of FIDO2/U2F compliant services, including popular cloud, email, and social platforms.
- Backup recommended – To ensure continuous access, register a backup Clife security key as a spare in case your primary key is lost.
2. Delegate only the authority the task needs
Define which data and actions an agent needs, then grant no more than that. Keep permissions appropriate to the intended audience and task; carry authorization context across calls to downstream agents and tools; and attenuate delegated rights so a later step does not silently gain the full authority of an earlier one. Modern protocols do not by themselves prevent overbroad roles or entitlement creep.
NIST’s August 27, 2026 article, “Back to the Future: Why Agentic AI Needs a Strong Identity Foundation,” discusses existing and emerging building blocks, including OAuth 2.0 and SPIFFE, as well as WIMSE, Identity Assertion JWT Authorization Grant work, Rich Authorization Requests, Transaction Tokens and the OpenID Foundation’s AuthZen. These are examples from an evolving ecosystem, not interchangeable solutions or a single NIST-endorsed architecture. Choose components based on the identity, delegation and policy requirements of the environment.
Rank #2
- Hardware-Rooted Security with PUF Technology – PUFido Drive Clife Key uses Physical Unclonable Function technology to generate a unique, hardware-based identity that cannot be duplicated, delivering stronger resistance against tampering and cyber attacks than conventional security keys.
- FIDO2 Certified Phishing-Resistant Protection – Fully compliant with FIDO2/U2F standards, enabling secure passwordless login and two-factor authentication to help protect accounts from phishing and credential theft.
- Security Key + Flash Drive in One Device – Combines a FIDO security key with a built-in USB flash drive, allowing you to carry files and a hardware authentication key together in a single compact device.
- Easy to Use & Portable – Compact USB-C design fits easily on a keychain or in a pocket. Simply plug in the Drive Clife Key to authenticate or access stored files with no extra software required.
- Universal Compatibility – Works with hundreds of FIDO2/U2F compatible services and supports Windows, macOS, Linux, iOS, Android, and other major platforms.
3. Treat tokens and secrets as lifecycle-managed credentials
Long-lived API keys and bearer tokens are risky because anyone who obtains them may be able to use or replay them. Credentials can also leak from plaintext configuration files, documents or logs. Store secrets in protected systems, scope them narrowly, set short lifetimes where practical, and establish processes to rotate and revoke them. Monitor for exposure and unexpected use.
Sender-constraining approaches such as DPoP can reduce some replay risks by making a token harder to use outside its intended context; they do not eliminate the need for narrow scope, safe storage or revocation. NIST finalized IR 8587, Protecting Tokens and Assertions from Forgery, Theft, and Misuse, on September 15, 2026. It provides implementation guidance on token protection and includes high-level AI considerations, but is not a comprehensive agent-security toolkit.
Rank #3
- Ultra-Compact FIDO2 Security Key - Plug-and-stay or carry on a keychain. This USB-A hardware security key offers portable, always-on protection for desktop and mobile use. (Item Size: 0.75 X 0.74 IN x 0.25 IN)
- USB-A Hardware Key for All Devices - Works with USB-A ports on PC, Mac, Android, and other laptop/notebook device. Enables secure, cross-platform login with FIDO2.0 passkey support.
- FIDO Certified Security Key - Meets FIDO and FIDO2 standards. Works with Google, Microsoft, GitHub, Dropbox, and more. Please check service compatibility before purchase.
- Passwordless Login with Passkey - Supports passkey login via WebAuthn and CTAP2. Enjoy password-free sign-ins where supported. Not all websites or services currently support passkeys.
- Advanced Multi-Factor Authentication - Offers 200 FIDO2 passkey slots and 50 OATH-TOTP slots. Strong, flexible 2FA/MFA support across various apps and authentication platforms.
4. Bound what the agent can reach at runtime
Limit the agent’s access to tools, files, APIs and data to what its task requires. Where appropriate, isolate execution in a hardened environment or tightly controlled container, and monitor tool use and authorization changes. Containment can limit the consequences of misuse; it does not guarantee protection against prompt injection or make an agent’s actions trustworthy.
Pay particular attention to local agents that run with a user’s credentials. They can impersonate that user and inherit broad access, weakening attribution and complicating centralized governance. A separate workload identity and constrained runtime are more auditable than letting the agent operate as the logged-in employee.
Rank #4
![Thetis Pro FIDO2 Security Key Passkey with Complex Pin [PinPlex], Hardware Device Supports USB A, Type C &NFC, TOTP/HOTP Authenticator APP, PIV Certificates, FIDO 2.0 Two Factor Authentication 2FA MFA](https://m.media-amazon.com/images/I/31ugrBqG6SL._SL160_.jpg)
- Dual USB-A and USB-C Security Key – Features both USB-A and USB-C connectors for seamless compatibility across desktops, laptops, and tablets. Supports plug-and-stay use or keychain carry.
- NFC-Enabled for Mobile Access – Built-in NFC allows fast, wireless authentication with Android and iPhone devices. Ideal for mobile logins and on-the-go security.
- FIDO Certified for Strong Authentication – [CHECK COMPATIBILITY before purchase] Fully compliant with FIDO2 and FIDO U2F standards. Works with major platforms like Google, Microsoft, GitHub, and Dropbox.
- Passwordless Login with PinPlex – Supports secure passkey login via WebAuthn and CTAP2 with added protection from PinPlex, a complex PIN system that enhances physical security.
- Multi-Layer Authentication Support – Includes PIV certificates and supports both TOTP and HOTP for strong 2FA/MFA coverage across enterprise and consumer apps.
5. Use human approval selectively
Require review for actions whose consequences warrant a person’s judgment, and preserve an audit record of the approval and resulting action. Do not make a prompt the primary security boundary: frequent requests can create consent fatigue, and a user may approve without meaningful review. Pair risk-based approval gates with narrow permissions, policy enforcement and runtime monitoring. NIST’s reviewed materials do not prescribe a universal approval threshold; organizations must set one according to the action’s impact and their risk tolerance.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Put the controls into an operating process
- Inventory agents and owners. Record each agent, its responsible operator, purpose, environment, tools and data access. Include agents embedded in workflows and locally run agents, not only centrally managed services.
- Map each task to permissions. Identify the specific actions and resources needed, then replace broad inherited roles with task-appropriate scopes and delegated authority.
- Issue and protect credentials. Avoid shared employee logins and persistent secrets in files or logs. Use protected storage, credential lifetimes and revocation procedures suited to the integration.
- Constrain execution and observe behavior. Set tool and data boundaries, isolate execution where appropriate, and retain records that let teams review actions, access decisions and changes in authorization.
- Set review and retirement rules. Decide which consequential actions need human approval, who can approve them, and how to suspend or retire an agent identity when its owner, task or environment changes.
When choosing an implementation, compare how clearly it attributes actions, limits and attenuates privileges, protects and revokes credentials, constrains runtime access, exposes policy decisions and supports meaningful human review. The sources cited here provide these as decision dimensions, not comparative test results or a vendor ranking.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesWhat current NIST guidance does—and does not—establish
NIST’s work is active, but the documents serve different purposes and should not be conflated with a finished agent-security compliance regime.
- Agent identity project: NIST’s National Cybersecurity Center of Excellence (NCCoE) published a concept paper on software and AI agent identity and authorization on February 5, 2026. Its public-comment period ended April 2, 2026. The NCCoE resource hub, accessed October 4, 2026, describes implementation-oriented guidance and an SP 1800-series practice guide with example implementations, architectures, build details and lessons learned as planned or ongoing. The hub reports over 600 responses to the concept paper; that is a count of responses, not unique people, organizations or deployments.
- Security considerations: NIST’s May 18, 2026 summary of responses to a separate AI-agent security request for information reports widespread agreement among commenters that agents bring novel threats and that foundational cybersecurity practices need adaptation. This summarizes commenters’ views; it is not a population-wide percentage or evidence that every respondent agreed. NIST’s summary and CAISI announcement identify concerns including indirect prompt injection, insecure or poisoned models, specification gaming and harmful actions.
- Standards and research initiative: NIST’s AI Agent Standards Initiative, created February 17, 2026 and updated August 14, 2026, describes work across industry-led standards, community-led protocols, and research into agent authentication, identity infrastructure and security evaluation. It signals ongoing development, not a settled set of requirements.
- Human digital identity: SP 800-63-4, finalized July 31, 2025, covers identity proofing, enrollment, authenticators, authentication, federation and assertions for users interacting with government information systems. It supersedes SP 800-63-3, but does not by itself define agent identity.
- Organizational AI risk: The voluntary AI Risk Management Framework (AI RMF) 1.0 was released January 26, 2023. It can help frame organizational AI risk management, but it is broader than agent identity and authorization; NIST’s framework page says revision is underway.
For enterprise teams, the practical takeaway is to apply established identity, access, credential and monitoring controls now while treating agent-specific standards as developing work. NIST’s publications are useful inputs to that work, not a substitute for defining the organization’s own permissions, approval rules and operating responsibilities.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
