AI Security News
Latest articles and news about AI Security on AXL Media.
Latest Articles
- Irregular Named Israel’s Most Promising Startup for 2026 Following Partnership with Anthropic and OpenAI
Published: Apr 28, 2026
Section: Israel
Tel Aviv-based security lab Irregular has claimed the top spot on Calcalist and CTech’s annual list of the most promising startups, fueled by its specialized focus on controlling a...
- Suzu Labs Acquires Emulated Criminals to Launch AI Driven Continuous Adversarial Operations Practice
Published: Apr 27, 2026
Section: AI in Business
Las Vegas based Suzu Labs has announced the acquisition of Emulated Criminals, a specialized cybersecurity firm focused on red teaming and adversary emulation. The merger establish...
- Researchers unveil ProbsCut to resolve the accuracy and robustness trade-off in neural networks
Published: Apr 25, 2026
Section: Technology
A research team led by Yun Li has developed ProbsCut, a new adversarial training method that utilizes bias-variance decomposition to improve deep neural network security. By implem...
- Anthropic’s Mythos AI Forces Global Security Pivot as Automated Vulnerability Discovery Reaches Machine Speed
Published: Apr 24, 2026
Section: Technology
The limited release of Anthropic’s Claude Mythos model has triggered a strategic shift among Chief Information Security Officers (CISOs) toward autonomous defense systems. Cybersec...
- Cyera Acquires AI Startup Ryft for $130 Million to Secure Autonomous Agent Data Access
Published: Apr 24, 2026
Section: Companies & Industry
Data security unicorn Cyera has announced the acquisition of Israeli startup Ryft for an estimated $100 million to $130 million to enhance its "agentic AI" security capabilities. T...
- Anthropic Endorses EPSS Model to Tackle AI-Accelerated Wave of Machine-Speed Software Vulnerabilities
Published: Apr 23, 2026
Section: Technology
Anthropic has recommended the Exploit Prediction Scoring System (EPSS) as a vital triage tool to manage an impending surge in software flaws discovered by its AI system, Mythos. As...
- Anthropic’s Mythos Model Identifies 271 Security Flaws in Latest Mozilla Firefox Source Code
Published: Apr 22, 2026
Section: Technology
Mozilla reports that Anthropic’s Mythos Preview AI model successfully pre-identified 271 security vulnerabilities in the source code of Firefox 150. Firefox CTO Bobby Holley sugges...
- Anthropic’s Claude Mythos Breakthrough Challenges Foundational Assumptions of Israeli Cybersecurity Sector
Published: Apr 17, 2026
Section: Israel
A technological leap by Anthropic, which allowed a general AI reasoning engine to autonomously identify decades-old vulnerabilities in hardened systems like OpenBSD, is forcing a s...
- Enterprise AI agents compromised: Microsoft and Salesforce address critical ‘ShareLeak’ and ‘PipeLeak’ data theft flaws
Published: Apr 17, 2026
Section: Science & Tech
Security researchers have uncovered high-severity prompt injection vulnerabilities in Microsoft Copilot Studio and Salesforce Agentforce that allow attackers to hijack AI behavior...
- Critical 'MCPwn' vulnerability in nginx UI triggers emergency warnings as active exploitation begins
Published: Apr 17, 2026
Section: Technology
Security researchers have identified a critical vulnerability in the open-source nginx UI management tool that allows unauthenticated attackers to achieve full server compromise wi...
- Cybersecurity landscape 2026: AI-powered threats evolve into machine-speed espionage and supply chain hijacking
Published: Apr 17, 2026
Section: Science & Tech
The 2026 Threat Detection Report reveals that while AI remains a primary tool for defense, it has dramatically lowered the entry barrier for sophisticated cyber attacks by automati...
- Anthropic Launches Project Glasswing to Secure Critical Infrastructure After New AI Model Uncovers Thousands of Zero-Day Vulnerabilities
Published: Apr 11, 2026
Section: Science & Tech
Anthropic has unveiled Project Glasswing, a defensive consortium including Amazon, Google, and NVIDIA, to combat a "profound shift" in cybersecurity. The initiative follows the dis...