Common Vulnerabilities And Exposures
PulseAugur coverage of Common Vulnerabilities And Exposures — every cluster mentioning Common Vulnerabilities And Exposures across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
Anthropic proposes AI jailbreak severity scale modeled on CVEs
Anthropic has introduced a new framework called the Cyber Jailbreak Severity (CJS) scale, designed to classify the danger of AI jailbreaks similarly to how CVEs grade software vulnerabilities. This system ranges from CJ…
-
BERT models compared for CVE-to-CWE mapping in cybersecurity research
A new research paper explores the effectiveness of different BERT models for mapping Common Vulnerabilities and Exposures (CVE) to Common Weakness Enumeration (CWE) categories. The study compares multi-class and multi-l…
-
New security pipeline audits 8,764 MCP servers, catching vulnerabilities
A new security auditing pipeline called Sentinel has been developed to address the significant number of CVEs filed against MCP servers. The pipeline employs a multi-layered approach, including static analysis, behavior…
-
AVE clarified: AI agent weaknesses classified like CWE, not CVE
The author clarifies that AVE (Agent Vulnerability Enumeration) is more akin to CWE (Common Weakness Enumeration) than CVE (Common Vulnerabilities and Exposures). Unlike CVEs, which identify specific flaws in particular…
-
CISA maintains CVE list of cybersecurity flaws
The Cybersecurity and Infrastructure Security Agency (CISA) maintains a standardized list known as Common Vulnerabilities and Exposures (CVE). This catalog documents publicly disclosed cybersecurity flaws found in both …
-
AI cuts zero-day exploit time to 8 hours
AI is significantly accelerating the exploitation of software vulnerabilities, with the average time from a CVE being published to its active exploitation now reduced to just 8 hours. This rapid timeline, tracked by the…
-
Syscall-based HIDS generalization to unseen CVEs shows mixed results
Researchers explored the generalization capabilities of system-call based Host Intrusion Detection Systems (HIDS). They investigated whether a HIDS trained on normal behavior for specific Common Vulnerabilities and Expo…
-
Microsoft uses AI to streamline record-breaking Patch Tuesday security updates
Microsoft's June Patch Tuesday update addressed a record 206 CVEs, with AI playing a role in streamlining the process. The company is leveraging AI to enhance its security vulnerability management, making the traditiona…
-
VLLM AI inference engine suffers 36 unpatched vulnerabilities
VLLM, an open-source AI inference engine, has a significant number of vulnerabilities, with 36 reported CVEs. Of these, 14 are classified as critical or high severity, and one has a maximum CVSS score of 10. A large maj…
-
AI Industrializes Vulnerability Discovery, Straining Cybersecurity Remediation
The traditional approach to cybersecurity, focused on identifying and fixing vulnerabilities, is becoming unsustainable due to the sheer volume of discovered flaws. AI is accelerating vulnerability discovery, outpacing …
-
Open-source software plagued by numerous CVE disclosures
Multiple vulnerabilities have been disclosed this week within open-source software projects. The disclosures highlight a widespread issue affecting various components of the open-source ecosystem. This situation undersc…
-
Ubiquiti Patches Critical Flaws; FreeBSD Earns Security Accolades
This week's security news highlights Ubiquiti's release of a bulletin detailing fixes for six security vulnerabilities, including two rated as critical. Separately, FreeBSD has been recognized for its robust security pr…
-
Critical 7-Zip vulnerability risks code execution on millions of machines
A critical security vulnerability, rated 8.8 on the CVE scale, has been discovered in the widely-used open-source file archiving utility 7-Zip. This flaw allows for arbitrary code execution simply by opening a specially…
-
Prompt engineering guide details structured data extraction from advisories
This tutorial details a method for extracting structured data from unstructured text, specifically focusing on cybersecurity advisories. It outlines a process using the OpenAI API, Pydantic for schema definition and val…
-
AI tool MOAK autonomously exploits new security vulnerabilities
MOAK, also known as Mother of All KEVs, is a new AI-powered tool designed to rapidly exploit newly discovered security vulnerabilities. This agentic workflow can autonomously identify and exploit CVEs as soon as they ar…
-
AI security tools may hallucinate vulnerabilities from training data
Large language models used for AI-assisted vulnerability discovery can falsely present information from their training data as novel findings. This occurs because LLMs cannot distinguish between recalling information ab…
-
Mythos AI model "discovers" vulnerability already in training data
The AI model Mythos has "discovered" a vulnerability in FreeBSD that was already present in its training data. This situation raises concerns about the reliability and safety of AI models trained on vast datasets, as th…
-
CyberSecQwen-4B: Small, specialized model offers local defense for cybersecurity
A new, specialized language model named CyberSecQwen-4B has been developed for defensive cybersecurity tasks. This model is designed to be small, runnable locally, and handle sensitive data without needing external APIs…
-
Security engineer warns Claude.md needs more than Karpathy's four lines
A security engineer discusses the limitations of Karpathy's four lines of code for improving AI model correctness, highlighting that these lines do not address security vulnerabilities. The author points out that despit…
-
AI agents reconstruct vulnerabilities from Linux and Windows binaries
Researchers have developed agentic pipelines for vulnerability discovery in software binaries. Patch2Vuln focuses on Linux distribution binary patches, successfully identifying security-relevant functions in half of tes…