Guidelines  ·  2026-09-17

NIST finalizes IR 8587 'Protecting Tokens and Assertions from Forgery, Theft, and Misuse' — explicitly flags AI-agent authorization as an open gap requiring further standards

GuidelinesMedium impactUnited States
NIST, with CISA support, finalized NIST IR 8587, 'Protecting Tokens and Assertions from Forgery, Theft, and Misuse,' providing implementation guidance for protecting digitally-signed tokens/assertions (SSO, API access) from post-authentication misuse. The publication builds on NIST SP 800-53 updates and responds to Executive Order 14306. Coverage published Sept 16, 2026 (CSOonline) reports that AI-agent actions are explicitly out of scope of the current guidance, but NIST states the same token-hardening principles should apply to AI agents while acknowledging that agentic-AI access risk 'creates additional IAM challenges that require further guidelines and, in some cases, new or expanded standards and protocols' — i.e., NIST/CISA flag agent authorization as unresolved and under active work via the parallel NIST AI Agent Standards Initiative.
This is a general (non-AI-specific) identity/token security standard, so it sits at the edge of scope, but it directly shapes near-term AI-agent security posture: it's the first finalized NIST guidance to explicitly name AI-agent token/authorization handling as an acknowledged standards gap, and enterprises deploying agentic AI are being told to apply token-hardening controls now while NIST develops agent-specific extensions.
Apply NIST IR 8587's token-lifecycle hardening (short-lived tokens, continuous monitoring, tighter post-auth controls) to AI-agent-held credentials as an interim measure; track the NIST AI Agent Standards Initiative for forthcoming agent-specific identity/authorization standards.
NIST News: NIST Finalizes Guidelines on Protecting Online Identity and Access Tokens From MisuseCSOonline: AI agent authorization risks remain a gap in new NIST-CISA token security guidance
See this in the live feed Explore related AI security and governance findings — updated every morning.
Open the feed →