Security researchers have uncovered four critical vulnerabilities in the Dify AI platform that collectively affect over 1 million AI applications. Dubbed “DifyTap,” these bugs enable attackers to execute arbitrary code, steal sensitive data, hijack administrative accounts, and compromise entire AI w
A sophisticated malicious AI agent skill successfully evaded all major security scanners and infected approximately 26,000 AI agents before detection. The compromised skill exploited trust mechanisms in AI agent marketplaces, demonstrating a new attack vector targeting autonomous AI systems. This in
Attackers are exploiting AI memory mechanisms to inject malicious prompts, exfiltrate sensitive data, and manipulate model behavior in Microsoft’s AI-powered applications. This emerging threat vector targets the persistent memory features in AI assistants, allowing adversaries to poison conversation
Security researchers have disclosed a critical zero-click exploit chain targeting Google Pixel 10 devices that leverages a vulnerability in the Visual Processing Unit (VPU) driver to achieve full root access without user interaction. The attack chain, discovered during internal security testing, com
Security researchers have disclosed AutoJack, a novel attack vector that exploits autonomous AI agents to achieve remote code execution (RCE) on host systems through maliciously crafted web pages. When an AI agent visits a compromised website, attackers can hijack the agent’s decision-making process
Chief Information Security Officers (CISOs) are experiencing unprecedented pressure from two converging forces: an intensifying threat landscape and the rapid integration of artificial intelligence into both attack vectors and defense strategies. Organizations are restructuring cybersecurity teams,
AutoJack is a critical vulnerability affecting AI agents that autonomously browse web content. By crafting a malicious webpage with embedded instructions, attackers can achieve remote code execution (RCE) on the host system running the AI agent. This attack exploits the trust boundary between natura
The Department of Justice has seized two prominent deepfake pornography websites, CFAKE and SOCFAKE, marking the first major enforcement action under the TAKE IT DOWN Act. These platforms enabled users to create non-consensual intimate imagery using AI technology, affecting thousands of victims glob
Security researchers discovered that AI coding assistants can be tricked into generating malicious code through seemingly innocent “fix this code” prompts—no sophisticated jailbreaking required. The vulnerability, demonstrated with a system dubbed “Fable 5,” raised concerns among federal authorities
A critical vulnerability dubbed “SearchLeak” has been discovered in Microsoft’s Copilot AI assistant that allows attackers to exfiltrate sensitive data through a single-click exploit. By manipulating Copilot’s search integration features, malicious actors can craft prompts that extract confidential