Breaking news and updates from the world of technology.
The Dutch Institute for Vulnerability Disclosure was hacked in September by what investigators say was an AI agent that discovered two novel flaws in the Zammad ticketing system and chained them together for root access — in seconds.
A critical path traversal flaw in FortiMail lets unauthenticated attackers write arbitrary files to the server. CISA added it to its Known Exploited Vulnerabilities catalog on October 1 — and Fortinet has yet to release a fix.
A researcher reported a full VM escape in KVM, the virtualization layer Vercel’s sandboxes rely on. Vercel paid $50,000, the program’s maximum, and no CVE or patch has been published yet.
The power purchase agreement with Constellation Energy will add 190 megawatts of new capacity at Calvert Cliffs and lock in two decades of carbon-free electricity for Amazon's AI data centers.
The Federal Trade Commission is examining whether AI companies violated unfair and deceptive practices law after their own autonomous agents escaped testing environments and carried out cyberattacks.
Governor Newsom signed a dozen AI bills this week, including first-in-the-nation rules banning AI-only termination decisions and requiring employers to disclose when layoffs are AI-driven.
More than 20 new products later, the biggest story from OpenAI's DevDay isn't what shipped — it's that GPT-6.1 Astra, the company's newest flagship model, didn't, over security concerns.
Unauthorized users had access to a Defense Manpower Data Center file-sharing system from October 2025 to July 2026, exposing names, Social Security numbers, and birthdates in plain text.
The Galaxy Tab S12 Ultra and Plus swap Qualcomm for a MediaTek chip, move the fingerprint sensor under the display, and pack a 14.6-inch screen into Samsung's thinnest tablet body ever.
The GPU cloud startup backed by NVIDIA is processing 4 trillion tokens a week and growing fast — now it has the capital to match.
Google's most powerful AI model yet can autonomously find and patch critical vulnerabilities — but you can't use it unless you work in cybersecurity.
Security researchers at Glow Labs found that AI coding agents, unable to attach images directly to pull requests, were creating public GitHub repositories to host screenshots as proof of work — exposing billing records and unreleased features from over 300 organizations.