On August 18, the "2026 IDC China Artificial Intelligence and Data Summit," hosted by IDC, was grandly held in Beijing, bringing together numerous industry elites and technology leaders. With its solid technical foundation and mature, proven industry solutions in AI security, NSFOCUS was named to the "2026 IDC China AI...
Blog
AI Security Incident Case: Encrypted Reasoning Blocks of Proprietary LLMs Can Be Stolen via Cross-Model Replay
Overview On August 10, 2026, MATS Research, the ELLIS Institute Tübingen, the Max Planck Institute for Intelligent Systems, and other institutions jointly published the paper Stealing Reasoning Traces from Proprietary LLM APIs. The research reveals a common architectural flaw in the reasoning model APIs of three major AI providers, Anthropic,...
Microsoft’s August Security Update of High-Risk Vulnerability Notice for Multiple Products
Overview On August 12, NSFOCUS CERT monitored that Microsoft released its August security update patch, fixing 421 security issues. These affect widely used products such as Windows, Microsoft Office, Azure, Visual Studio Code, Microsoft Exchange Server, etc., which include high-risk vulnerability types such as privilege escalation and remote code execution....
NSFOCUS AI CyberGym Level 1 Technical Report
Abstract NSFOCUS's self-developed intelligent offensive-and-defensive system NSFOCUS AI achieved a vulnerability-reproduction success rate of 93.6% on the CyberGym Level 1 benchmark. Behind this result lies the deep integration between NSFOCUS AI and Zhipu's GLM-5.2 large model: by formalizing the offensive-and-defensive expertise accumulated by NSFOCUS's red team into computable workflows, the...
NSFOCUS Monthly APT Insights – April 2026
Regional APT Threat Situation In April 2026, the global threat hunting system of Fuying Lab detected a total of 29 APT attack activities. These activities were primarily concentrated in regions including East Asia, South Asia, Eastern Europe, North America, and the Middle East, as shown in the figure below. Regarding...
AI Security Incident Case: AISI Reveals AI Agents Autonomously Attacking Real People and Systems During Security Testing
Overview In August 2026, the UK AI Security Institute (AISI) disclosed a startling security incident: during routine cybersecurity capability evaluations, Anthropic's Mythos 5 and OpenAI's GPT-5.6 Sol models, without receiving any explicit attack instructions, autonomously decided to launch unauthorized actions against real open-source project maintainers and external systems. Their specific...
