A counterfeit Perplexity AI Chrome extension was discovered intercepting user search queries and address bar inputs, redirecting traffic through attacker-controlled domains. The malicious extension mimicked the legitimate Perplexity AI service, exploiting user trust to harvest browsing data and pote
A critical security vulnerability in Amazon’s Q AI coding assistant for Visual Studio Code could allow attackers to steal AWS credentials through malicious code suggestions. The flaw enables credential exfiltration when developers accept and execute compromised code snippets generated by the AI assi
Chinese AI company Zhipu AI has released GLM-5.2, claiming performance comparable to Anthropic’s Claude in identifying security vulnerabilities. This development signals a significant shift in the AI-powered security landscape, as state-backed Chinese models potentially match Western capabilities in
Secret Service agents are reportedly avoiding government-issued devices in favor of personal phones, raising significant operational security (OPSEC) concerns. This practice, while potentially driven by usability or security concerns, creates data exfiltration risks, circumvents monitoring capabilit
A critical security vulnerability in Amazon Q Developer allows maliciously crafted Git repositories to execute arbitrary code and steal AWS credentials when developers clone and interact with compromised repos. The flaw exploits Amazon Q’s code analysis features, turning a productivity tool into an
Contrary to apocalyptic predictions, AI is generating new entry-level cybersecurity opportunities rather than eliminating them. As organizations deploy AI-powered security tools, they need human analysts to validate findings, tune systems, and make contextual decisions. The shift creates demand for
North Korean threat actors have deployed macOS.Gaslight, a sophisticated malware strain designed to deceive security analysts by manipulating AI-powered analysis tools. The malware employs anti-analysis techniques that specifically target machine learning detection systems, generating false benign i
Recent research exposes critical flaws in Large Language Model (LLM) defense architectures through “role confusion” attacks, where prompt injections manipulate AI systems into misinterpreting instructions and security boundaries. The paper demonstrates how current safeguards fail when attackers expl
Google has released Gemini 3.5 Flash with computer use capabilities that enable autonomous AI agents to interact directly with systems, applications, and interfaces. While these features unlock powerful automation possibilities, they introduce significant security risks including unauthorized system
Artificial intelligence is fundamentally disrupting cybersecurity operations by discovering vulnerabilities faster than traditional security frameworks can handle. Organizations face an unprecedented challenge: AI-powered tools are identifying thousands of potential weaknesses daily, overwhelming ex