Termsslug: vulnerability-discovery

Vulnerability Discovery

漏洞挖掘

Use of AI systems to automatically identify security flaws in software and infrastructure.

Curated Intelligence

9 items · Page 1/1
OpenAI 官方博客T1·Industry
58 pts
OpenAI Staffer Warns AI Policy Is Being Shaped by Urgent Crises

In a Harvard Vik-Bailey lecture, an OpenAI staffer said frontier AI systems have uncovered critical vulnerabilities in core internet infrastructure, and that such "Black October" episodes will recur, forcing AI policy to be made amid perpetual urgency. The author argues the goal should be preserving human agency and the capacity to ask new questions and build new things, not merely comfortable machine-provided lives.

IT之家 · 智能时代T2·Industry
58 pts
Cloudflare CEO Says AI-Driven Vulnerability Wave to Last Two Years

Cloudflare CEO Matthew Prince said on the Q3 earnings call that AI is rapidly surfacing hidden software vulnerabilities, and severe security threats are expected to keep emerging over the next two years. CISOs are responding by signing large security contracts and favoring integrated platforms over point solutions.

The DecoderT2·Models
86 pts
OpenAI Unveils GPT-6 Cyber to Find Exploits and Suggest Patches

OpenAI has launched GPT-6 Cyber, a model purpose-trained for vulnerability hunting that can find flaws, build working exploits, and propose patches. In targeted tests it surfaced 81 previously unknown critical Firefox vulnerabilities already patched by Mozilla, and in the Linux kernel reproduced 1,724 vulnerabilities with working exploits for 948—more than five times its predecessor.

TechCrunch AIT2·Models
72 pts
OpenAI Opens Security Model That Found Severe Ubuntu Flaw in Two Hours

OpenAI is giving cybersecurity defenders access to frontier models through its Daybreak Blue program, and one developer used GPT 5.6 Sol to surface a severe Ubuntu server vulnerability in about two hours; Debian has since patched it. The episode shows LLM-driven vulnerability discovery is now practical, while the piece argues the industry still lacks urgency for patching critical digital infrastructure.

The DecoderT2·Models
78 pts
GPT-5.6 and Gemini 4 Discover Critical PHP Vulnerability

OpenAI's GPT-5.6 Sol and Google's Gemini 4 Argon independently discovered a critical PHP remote code execution vulnerability affecting millions of websites. Verified by security firm Wiz, the finding highlights LLMs' capability in automated vulnerability discovery while sparking debate over AI lowering the barrier for attackers.

IT之家 · IT业界T2·Industry
58 pts
China Proposes Export Restrictions on AI Malware and Biotech Tech

China's Ministry of Commerce and MOST released a draft for public comment proposing to add 40 key technologies in AI, biomedicine, and photovoltaics to the export control list. Notably, AI-driven malware generation is proposed for prohibition, while AI-assisted vulnerability mining faces restrictions, aiming to prevent technology misuse and regulate international cooperation.

IT之家 · 智能时代T2·Industry
42 pts
Anthropic Opens Claude Mythos for Vulnerability Discovery

Anthropic launched a new cyber verification program, integrating previous initiatives to grant tiered access to its most powerful model, Claude Mythos, for vetted security teams. The initiative aims to leverage AI for vulnerability discovery and defense, having already identified over 100,000 bugs.

Google DeepMindT1·Models
78 pts
Google Launches Gemini 3.5 Flash Cyber to Power CodeMender Security Agent

Google introduces Gemini 3.5 Flash Cyber, a security-optimized model designed to counter the asymmetry where AI finds vulnerabilities faster than humans can fix them. Serving as the engine for the CodeMender agent, it performs hours-long deep code analysis and has demonstrated autonomous vulnerability discovery and patching in large-scale projects like Chromium.