AI Infrastructure
AI infrastructure explores GPUs, compute, training, inference and large-scale deployment. Articles follow platform evolution, architectural bottlenecks and the security implications of AI workloads.
Bifrost AI Gateway RCE Risk: Authentication Disabled by Default
CVE-2026-90898 in the Bifrost AI gateway enables unauthenticated remote command execution. The flaw is amplified by a design choice th…
JADEPUFFER Encrypts AI Weights with ENCFORGE: Phantom Ransom, Broken Payment Mechanism
The agentic ransomware group JADEPUFFER struck Langflow infrastructure with ENCFORGE, a payload that encrypts AI models without a func…
Plugin4Shell: First Supply-Chain Vulnerability in AI Agent Plugin Ecosystem
The Plugin4Shell vulnerability bypasses SHA pinning in plugins for four AI agents. Two vendors have patched; Microsoft and Google have…
Microsoft Fixes CVSS 10.0 Bug in Azure AI Foundry With Zero Customer Action
CVE-2026-85889 earned the maximum CVSS 10.0 score for a missing authentication flaw in Azure AI Foundry. Microsoft mitigated the issue…
Flowise: Critical RCE in CSV Agent, Component Removed from Codebase
CVE-2026-70477 with CVSS 9.8 in Flowise's CSV_Agent component: unauthenticated prompt injection leads to remote code execution. The ve…
AMD, Arm, and Nvidia: September 9 Patches for GPU and AI Inference Vulnerabilities
On September 9, 2026, three chipmakers released coordinated security advisories: flawed GPU drivers and bugs in Nvidia's Triton Infere…
LiteLLM: 9.6% of Exposed Instances Vulnerable to Auth Bypass and RCE
CISA adds CVE-2026-59822 to KEV catalog: roughly 9.6% of public LiteLLM instances accept default master key or no authentication, with…
ChatGPT's Isolated Containers Could Talk: Cross-Account Data Exfiltration
Check Point Research demonstrated a covert data-exfiltration channel between code-execution sandboxes belonging to different ChatGPT a…
OpenAI: Autonomous Agents Used an Abandoned German Wiki for Three Months
Roughly 18,000 posts left on DSEwiki between May and July 2026. The GET/POST bypass mechanism reveals a gap between security testing a…
CVE-2026-0768: Active Exploitation of Langflow, 360+ Attacks in Hours
The Langflow AI platform is under massive exploitation: over 360 attempts detected in hours leveraging critical vulnerability CVE-2026…
LLM Safety Rests on 50 Neurons: How Unit 42 Shatters the Myth of Distributed Alignment
Unit 42 research shows that safety mechanisms in large language models reside in less than 0.02% of neurons. The perturbation probing…
AI Honeypot: Real Attacks on LiteLLM and MCP Servers Reveal Three Patterns
Wiz Threat Research deployed AI/ML honeypots for 90 days and documented sustained, service-specific attacks. Three distinct patterns t…