Claude Hacked Real Companies by Accident
Anthropic says several Claude models breached three real organizations during security tests, thinking the live networks were part of a simulation.
Arun Kumar
Google Gives Robots a Faster Brain
Gemini Robotics ER 2 lets robots watch video, track their own progress, and team up with other machines in real time.
Freehand Bets AI Agents Can Audit Supply Chains
A San Francisco startup just raised $75M to let AI agents check giant supplier invoices and haggle over the difference, no human required.
Fish Audio's $52M Bet on AI Voices
A voice model that started on a single GPU now has 8 million users and a thorny consent problem to solve.
Sutskever's SSI Bets Big on Nvidia
After two quiet years, Ilya Sutskever's superintelligence lab breaks cover with a multi-billion-dollar Nvidia compute deal.
What's Buzzing This Week! (July 18-25, 2026)
Chip alliances, a surprise CPU, a $10.3B startup, and an AI model that learns to see, hear, and watch.
FLUX 3 Learns the World, Not Just Pixels
Black Forest Labs' FLUX 3 trains on images, video, and audio at once, betting one model can power both content creation and robots.
Etched's AI Chip Bet Hits $10.3B
Three Harvard dropouts built a chip for AI inference, and now investors are writing checks that value it at $10.3 billion.
AMD Bets $5 Billion on Anthropic
AMD will invest up to $5 billion in Anthropic and supply the chips to train and run Claude, deepening a chipmaker and lab alliance.
Nvidia's Vera CPU Chases a Bigger Market
Nvidia just shared benchmarks and details for Vera, its first CPU with a custom core, and the pitch isn't what you'd expect.