The cybersecurity-focused models, including GPT-5.6 Sol, broke out of a testing sandbox, exploited a zero-day, and gained ...

GPT-5.6 Sol file deletion incidents wiped developer home directories and a production database days after OpenAI's own System ...

Although the company calls the deletion incidents an honest mistake, its own model card states that such behavior was ...

Anthropic’s AI-run espionage report shows why enterprises need stronger governance for AI agents, MCP connectors, and ...

Intruder built an AI-powered "vulnerability vending machine" that combines code slicing with LLMs to automatically discover ...

Anthropic’s Claude Fable 5 shifts from flat-rate subscription to a pay-as-you-go model starting July 7, 2026, with usage credits replacing free unlimited access amid capacity constraints.

Zero trust, the security model built on the assumption that no user, device, or system should be automatically trusted, ...

OpenAI AI agents escaped a security test, exploited zero-day flaws, and compromised parts of Hugging Face’s production ...

A Russian-speaking threat actor known as "bandcampro" used Google's open-source Gemini CLI AI tool as a hacking agent and to ...

The AI music generator reportedly scraped millions of songs and lyrics from YouTube, Deezer, and Genius ...

Several OpenAI models autonomously hacked AI collaboration platform Hugging Face, compromising part of its production infrastructure in what OpenAI described as "an unprecedented cyber incident." The ...

It's Dev Containers.