Topic
91 articles
GoogleGTIG's AI-era report: CVE disclosures doubled in 2026, 141 bugs exploited by August vs 127 in 2025, and 50% of AI-discovered flaws lead to RCE, per Google.
Agentic AIOpenAI dismantled a coordinated campaign that weaponized cross-model replay queries across 15,000 accounts to illicitly extract encrypted chain-of-thought traces.
Threat IntelAn unauthenticated RPyC service in LightLLM visual_only deployments enables remote attackers to execute arbitrary code on GPU clusters via pickle deserialization.
Deep DiveAn architectural vulnerability in Model Context Protocol (MCP) OAuth token passthrough allows malicious tool servers to impersonate users and hijack enterprise APIs.
MicrosoftMicrosoft mitigates maximum-severity CVSS 10.0 flaw CVE-2026-70352 in Azure AI Language Authoring, preventing unauthenticated remote tampering with enterprise AI models.
MicrosoftMicrosoft discloses high-severity flaw CVE-2026-85917 in Azure AI Foundry allowing unauthenticated attackers to execute SSRF and elevate privileges over the network.
LinuxHigh-severity flaw CVE-2026-44222 in vLLM allows remote unauthenticated users to crash distributed GPU inference clusters via crafted multimodal tokens.
Threat IntelVulnerability CVE-2026-67620 in Flowise AI workflow builder allows attackers to bypass SSRF deny-lists and extract cloud instance metadata credentials from OCI and Alibaba Cloud.
MicrosoftMicrosoft patches maximum-severity CVSS 10.0 flaw CVE-2026-85889 in Azure AI Foundry enabling unauthenticated remote network privilege escalation.
Agentic AICISA adds critical Langflow AI vulnerabilities CVE-2026-0770 and CVE-2026-55255 to KEV catalog as threat actors weaponize agentic pipeline execution flaws.
Threat IntelAnthropic threat intelligence reveals nation-state groups, including Midnight Blizzard, attempted to weaponize Claude to automate malware obfuscation and EDR evasion.
Active ExploitCISA issues an emergency directive on CVE-2025-62593, warning that unauthenticated code injection in Ray AI distributed computing clusters allows adversaries to seize GPU clusters and training models.
AWS CloudCISA issues urgent warning on CVE-2026-64849, a critical SSRF defect in open-source MLflow model registries allowing attackers to steal cloud IAM metadata and poison AI pipelines.
Gov / DefenseA joint cybersecurity advisory from CISA, NSA, and FBI alerts defense contractors and cloud vendors to sophisticated state-sponsored AI model extraction campaigns.
Active ExploitVulnerability researchers observe mass automated scanning and exploitation of Langflow AI visual frameworks, exfiltrating OpenAI and AWS secrets while deploying botnets.
Deep DiveWiz Research discovered that 10% of internet-facing LiteLLM AI proxy gateways accepted the default documentation admin key sk-1234, leaking upstream model keys and cloud IAM tokens.
CISA added 15 CVEs to the KEV catalog in six working days, including exploited flaws in LiteLLM and Kestra — AI tooling that most asset inventories never see.
Deep DiveRegulation (EU) 2026/1744 pushed Annex III high-risk obligations to December 2027, but Article 12 logging, Article 14 oversight and Article 15 hardening are engineering work you cannot start late.
Agentic AIGravitee says 88% of organisations had a confirmed or suspected AI agent incident, while 82% of executives trust their policy. Both figures, one population.