Google’s Gemma 4 models now use quantization-aware training to shrink their memory footprint without sacrificing quality — a meaningful step forward for running AI on phones and laptops.
Leaks suggest OnePlus is developing smartphone displays with refresh rates that rival high-end gaming monitors — a bold move that raises questions about real-world utility versus spec-sheet bragging rights.
The Great American AI Act is a 269-page bipartisan bill that would regulate frontier AI models, fund open-source security, and preempt state AI laws. It’s ambitious, controversial, and could reshape the entire AI landscape.
A breach at Nelnet Servicing exposed Social Security numbers and personal data for 2.5 million student loan borrowers. With loan forgiveness in the news, phishing attacks using this data are a serious and ongoing threat.
Cisco disclosed its seventh exploited SD-WAN zero-day of 2026. CVE-2026-20245 allows root-level command execution with no patch available yet. Here’s what organizations need to know and do right now.
AI-generated content creators are becoming nearly indistinguishable from human ones, raising questions about authenticity, manipulation, and the future of creative work online.
Researchers at the University of Toronto have developed a training-free approach to single-image diffusion models that achieves leading results without any neural network training — replacing hours of compute with direct mathematical computation.
Lauf’s new eElja electric mountain bike weighs just 37.7 pounds and rides like a traditional MTB — with a hidden 350W motor. WIRED gave it a perfect 10/10, though the $6,490 starting price keeps it in premium territory.
Password manager Dashlane disclosed that a brute-force attack targeting its device registration system compromised fewer than 20 user accounts. Here’s what actually happened and what it means for you.
