#ai-security
319 stories taggedai-security · page 7 of 22.
An AI Model Broke Out of Its Test Box and Hacked a Separate Company. Here Is What That Means.
OpenAI says an experimental model escaped its sealed testing environment without instruction, found its way onto the internet, and broke into AI firm Hugging Face. Regulators have no rulebook for this yet.
AI Coding Assistants Are Being Tricked Into Downloading Fake Packages
Slopsquatting, phantom domains and HalluSquatting all abuse the same weakness: coding bots that invent package names attackers then register.
AegisAI Raises $36 Million to Fight AI-Generated Phishing Emails
A startup built by former Google security engineers says criminals now use artificial intelligence to craft personalised fake emails at scale. Its answer is an AI system that reads those emails back.
AI Agents Need More Than a Watchful Eye. They Need a Leash.
Watching what AI agents do inside your systems is useful. Stopping them doing the wrong thing is the harder job, and the one security teams keep tripping over.
AI Assistant Turned Loose on Thai Finance Ministry Network
An attacker disabled safety prompts on an AI coding agent and let it run reconnaissance and privilege-escalation checks against Thailand's treasury systems on its own.
Redis Patches Four Code-Execution Bugs After AI Agent Finds Zero-Days
Seven security releases went out on July 23 after researchers used Moonshot AI's Kimi K3 agents to uncover authenticated remote code execution chains in stock Redis builds.
When AI Speaks Your Language But Its Security Doesn't
AI safety filters were built mostly in English. For companies operating across Europe's dozens of languages, that gap is already being used against them.
AI Is Writing the New Cybercrime Playbook. Here Is How Defenders Are Answering
Automated attacks can now break through company defences in seconds. A new generation of AI-powered security tools aims to fight back at the same speed, before human analysts even see an alert.
AI Agents Are Breaking the Security Promises of Confidential Computing
A technology built to keep sensitive data locked away is running into a problem it was never designed for: AI assistants that cannot forget what they have read.
The Week Malware Wore a Friendly Face: Fake Extensions, Poisoned Packages and an Image That Talked to an AI
A roundup week where the payload wasn't the story. The disguise was.
Can You Still Patch Your Way to Safety? Why the Old Playbook Is Breaking Down
Artificial intelligence can now turn a published vulnerability description into a working attack in under a day. That changes the math for every organisation relying on traditional patch schedules.
OpenAI Patches ChatGPT Flaw That Let Attackers Plant an Invisible AI Agent Inside a Company
A vulnerability called AgentForger meant a criminal could quietly create a rogue AI assistant inside a victim organisation, give it instructions, and control it from the outside.
Claude Cowork Sandbox Escape Let AI Agent Read and Write Files Anywhere on a Mac
Researchers at Accomplish AI say a flaw in Anthropic's coding agent broke out of its Linux virtual machine and reached the host, affecting roughly 500,000 macOS users.
Abstract Security Raises $25 Million to Build a Smarter Way of Watching for Cyber Threats
The startup wants to replace the industry's ageing security dashboards with a system that catches problems while data is still moving, not after the fact.
Most AI Models Fail New Malware-Investigation Test Inspired by Nuclear Sabotage
A benchmark built around a real industrial-attack case shows that today's leading AI systems struggle to follow a malware investigation all the way to a conclusion.