By failure type

Incidents by type, October 2026 against September 2026 · as of 21 October 2026
TypeOctober 2026September 2026
Prompt injection43
Data exposure13
Rogue action13
Oversight gap21
Agent-enabled attack11

By class

Incidents by class, October 2026
ClassIncidents
Security exploit9

Every incident in October 2026

  1. LangAlpha agent's unguarded Bash and code-execution tools let web content trigger RCE and vault-secret theftginlix-ai (LangAlpha, open-source project) · Prompt injection · High · Vendor tooling
  2. IntentKit agent executor lets owner-only tools run without the human-in-the-loop approval gateCrestal Network (IntentKit project) · Oversight gap · High · Vendor tooling
  3. PraisonAI's AICoder agent let prompt-injected chat messages write files and run shell commands as root, in a batch of 19 disclosed flawsPraisonAI (MervinPraison/PraisonAI) users · Prompt injection · Critical · Vendor tooling
  4. PraisonAI's prompt-injection filter failed open while its email webhook accepted forged, unauthenticated agent triggersPraisonAI users (deployers of the open-source agent framework) · Prompt injection · High · Vendor tooling

Counted by the date each incident happened, from the Agent Failures Index. Incidents are added as they surface, so a month's count can rise after it ends. This month is still in progress.