AWS Security Blog · November 13, 2026

ARTEX campaign records: protect AI gateway credentials and verify reported outcomes

Why it matters

AWS’s MadPot decoys captured agent histories describing AI-key theft and cryptocurrency identity-check attacks. AWS attributes the tooling to ARTEX with moderate confidence; it did not observe or verify the off-sensor outcomes those agents claimed. The records describe exposed gateways, default credentials and leaked secrets funding further inference. One recorded deepfake liveness attempt repeatedly failed server-side checks, limiting claims of successful autonomous fraud.

My takeaway: Inventory public model gateways and inference endpoints, remove default keys, require authentication and patch exposed management services. Scan code, build artifacts and stored agent traces for secrets; rotate exposed credentials and review model usage and spend. Correlate AWS’s dated indicators with your own telemetry before declaring compromise. Keep attempted actions, agent-reported success and independently verified impact separate.
Keep exploring

More curated notes connected through Agent Security and AI Red Teaming.

OpenAI News · framework

Pacing model development in an era of cyber-critical capabilities

OpenAI says preliminary evidence that Astra may meet its Critical cybersecurity threshold led it to pause frontier reinforcement-learning work for two weeks and keep its largest planned run on hold. New safeguards include stronger workload and network isolation, continuous boundary testing, token-level monitoring that escalates suspicious tool activity, and broader alignment checks for deception, reward hacking, and unauthorized access.

OWASP GenAI Security Project · guide

OWASP Top 10 for Agentic Applications for 2026

OWASP's community guide organizes agentic-system risk into ten categories, including goal hijacking, tool misuse, identity and privilege abuse, memory poisoning, insecure inter-agent communication, cascading failures, and rogue-agent behavior. It provides a shared taxonomy and mitigation starting point rather than a certification checklist or evidence that a deployed system is secure.

OpenAI News · framework

A blueprint for democratic governance of frontier AI

OpenAI proposes a three-part U.S. frontier-AI governance model: harmonize emerging state safety laws into a federal baseline, strengthen CAISI as an evaluation and standards institution, and coordinate a broader resilience program. Proposed controls include severe-risk evaluations, transparency reports, independent audits, safety-incident reporting, model-weight security, whistleblower protection, and periodic technical assessments.