CISA and NIST Release Guidance to Secure Cloud Identity Tokens

The Cybersecurity and Infrastructure Security Agency (CISA) and the National Institute of Standards and Technology (NIST) have issued final guidance to protect cloud identity tokens and assertions from theft, forgery, and misuse. The report, Interagency Report 8587, provides technical standards to help federal agencies, cloud service providers, and their customers secure the tokens used in single sign-on (SSO), identity federation, and application programming interface (API) access.

Technical Requirements for Token Management

The new guidance introduces several strict controls for handling digital identities. To minimise the window of opportunity for attackers, the report states that access and identity tokens should be valid for no more than one hour. Authorization services and policy enforcement points are required to reject any expired tokens outright.

Key management protocols have also been tightened. For high-impact systems, signing keys must be rotated at least every 90 days, while a one-year rotation cycle is recommended for all other systems. For systems classified as moderate impact or higher, keys must be stored in hardware-backed or isolated storage. The guidance specifies that keys should never be stored persistently on the servers, virtual machines, or containers that use them.

High-impact systems are also required to sign tokens within an isolated execution environment. To prevent lateral movement by attackers, keys must be scoped to the narrowest possible boundary, ensuring a key valid outside a federally authorised environment cannot be used to sign tokens within one.

Additional security measures include the requirement that every token must carry an explicit audience field. Any access control mechanism that receives a token without this field must reject it. To protect user privacy, the guidance prohibits writing tokens or any personal data contained within them to system logs.

Mitigating Identity-Based Attacks and AI Risks

The guidance addresses the growing use of signed tokens by AI agents to access systems, data, and APIs. While NIST and CISA are still developing broader guidelines regarding the wider risks of AI agent access, this report confirms that current identity security standards apply to these automated entities.

The technical recommendations follow several high-profile security incidents. In 2020, a supply chain intrusion saw attackers compromise Active Directory Federation Services to forge SAML assertions, allowing them to bypass multi-factor authentication (MFA) at thousands of organisations. In a separate incident, the accidental exposure of a consumer signing key allowed foreign actors to forge tokens, resulting in the theft of more than 60,000 emails from a single agency.

“Identity is the new perimeter, and the tokens and assertions behind it are attractive targets for sophisticated adversaries,” said Chris Butera, CISA acting executive assistant director for cybersecurity. He noted that the guidelines provide a framework to harden token issuance, preventing a stolen credential from becoming a foothold across an entire enterprise.