Em 3 de fevereiro de 2026, uma grave vulnerabilidade (CVE-2026-25253, CVSS 8.8) foi divulgada no OpenClaw, um agente de IA de código aberto, permitindo execução remota de código. Isso levou à descoberta de 138 vulnerabilidades em 63 dias, com mais de 135.000 instâncias de OpenClaw publicamente expostas globalmente, muitas sem autenticação.
Cal.com closed its source code, fearing that AI turns sharing into a security risk as AIs can rapidly reverse-engineer and crack code. The article questions if open source is dead, criticizing the proposed solution for overlooking a fundamental principle.
Google has confirmed an AI system independently developed a zero-day exploit targeting two-factor authentication, signaling a new, automated threat in cybersecurity. This marks the first time defenders face an exploit born entirely from machine intelligence rather than human ingenuity.
OpenAI released GPT-5.5-Cyber to select participants in its Trusted Access for Cyber program, a variant fine-tuned for offensive and defensive security workflows. This specialized AI is hardened against system prompt injection and was briefed to federal agencies, covering capabilities like automated vulnerability discovery.
Breaking news reports indicate the financial sector is entering a new era of cybersecurity. This development involves leveraging AI against AI, a trend with significant global implications.
Elastic has launched the Security MCP App, an extension for the Model Context Protocol that embeds interactive security operations like alert triage and threat hunting directly into AI assistants. This tool provides visual dashboards within AI conversations, enabling SOC analysts to perform investigations and actions that synchronize in real-time with Elasticsearch and Kibana, reducing context switching.
The Model Context Protocol (MCP), a new standard for connecting AI agents to tools, is currently experiencing severe security flaws. A scan found that 36.7% of over 7,000 live MCP servers were vulnerable to SSRF, with hundreds lacking authentication or encryption. To address this, the AgentWarden CLI tool has been developed to scan MCP servers for real vulnerabilities.
The NSA has identified Model Context Protocol (MCP) as a national security concern due to its tool-calling architecture creating exploitable attack surfaces in AI-driven automation. This article focuses on how to operationalize the NSA's guidance to fix these security vulnerabilities.
This article highlights a critical security flaw where the AI code editor Cursor generated MD5 password hashes in 2026 due to outdated training data. This exposes user passwords to rapid cracking, and the author suggests simple fixes like bcrypt or argon2id.
This article highlights the security risk posed by the mismatch between the short lifespan of AI agents and the long lifespan of their credentials. It advocates for using a broker to issue task-scoped, short-lived credentials at an agent's spawn time to prevent credential sprawl and reduce the attack surface.
This article offers a practical guide for IT leaders on securing AI agents, addressing immediate operational requirements. It highlights the unique challenges of AI agent security compared to traditional applications due to their unpredictable behavior.
The cybersecurity landscape demands new approaches as traditional methods are insufficient against sophisticated threats. AI Cyber Defense offers a paradigm shift, utilizing intelligent systems to learn, predict attacks, and respond in real-time.
This article clarifies the distinctions between AI-enhanced SIEM, SOAR, and UEBA strategies in cybersecurity. It emphasizes understanding their different approaches to threat detection and and response for an effective security architecture.
AI has profoundly transformed cybersecurity with its speed and data analysis, yet it cannot fully take over due to its reliance on past patterns and inability to handle novel, human-driven threats. Human professionals provide essential intuition and critical thinking for unpredictable attack strategies that AI struggles to identify.
This paper introduces H-TechniqueRAG, a novel hierarchical Retrieval-Augmented Generation (RAG) framework designed to improve the annotation of adversarial techniques in Cyber Threat Intelligence (CTI) text. It addresses the limitation of flat RAG approaches by incorporating the inherent tactic-technique taxonomy of the MITRE ATT&CK framework through a two-stage retrieval mechanism.
Researchers from Ox Security uncovered critical vulnerabilities in MCP servers, affecting millions, which Anthropic refuses to fix at the protocol level. This architectural flaw allows malicious prompts to hijack tool execution and exfiltrate data from publicly exposed servers.
This article highlights the critical security flaw of missing audit trails in MCP servers, a standard protocol for AI agent-tool communication. It details how the absence of crucial features like request signing and caller identity verification has led to severe incidents, offering a practical checklist to mitigate these risks.
This article analyzes how Claude Opus 4.6 discovered over 500 zero-day vulnerabilities, including a 23-year-old Linux kernel bug, transforming LLMs into autonomous security research agents. It explores the technical mechanisms and DevSecOps implications of this AI-driven vulnerability discovery.
An autonomous AI agent details its experience of detecting and mitigating 368 SSH attacks within 24 hours of its deployment. Initially lacking security knowledge, the AI learned to implement tools like fail2ban and a file permission scanner to protect its system from intrusion attempts.
Em 7 de abril de 2026, a Anthropic anunciou o Project Glasswing, apresentando o modelo de IA Claude Mythos Preview, capaz de identificar milhares de vulnerabilidades zero-day em todos os principais sistemas operacionais e navegadores. Este modelo de fronteira demonstrou ser superior à detecção humana e automatizada, com profundas implicações para a cibersegurança.