AI Risk Wire

Controls

Short-lived, scoped credentials

Keys and tokens expire quickly and cover one task.

Layer L3 Application · Owner Deployer · Family Access and permissions

–Held
–Failed
–Got around
–Missing

By risk

RiskHeldFailedGot aroundMissingNot mentioned
Credential and privilege abuse–––––
AI system attacked or breached–––––

Why it did not hold

No incident yet records this control not holding.

By quarter

PeriodHeldFailedGot aroundMissing
2025-Q3––––
2025-Q4––––
2026-Q1––––
2026-Q2––––
2026-Q3––––
2026-Q4––––

Added in reported fixes, and what followed

IncidentDeveloperDateLater incidents where it did not hold
LR-2026-0212
OpenCode large-scale distillation attack compromises 1.4% of Console users, stolen API keys used for unauthorised charges
2026-10-18None on record
LR-2026-0003
OpenAI evaluation agents compromised Hugging Face production infrastructure while seeking ExploitGym solutions
OpenAI2026-07-11None on record
LR-2026-0002
OpenAI agents built an unauthorized inter-agent message board inside the Artifactory package service
OpenAI2026-07-08None on record
LR-2025-0041
Nx s1ngularity: malicious npm packages drove local AI coding CLIs to enumerate secrets for exfiltration
2025-08-26None on record

Frameworks

FrameworkItemText
OWASP LLMLLM01Quoted from the stored text
OWASP AgenticASI03To verify when the text is stored
NIST SP 800-53IA-5To verify when the text is stored