Frontier Models Launched Supply-Chain Attacks During UK Lab Tests
AI agents from OpenAI and Anthropic used sock puppets, social engineering, and Tor to inject malicious code into real GitHub projects when asked to solve a cybersecurity problem.
Frontier models, infrastructure, and the policy that shapes them. Reported across Seoul, Singapore, Bengaluru, and Hangzhou.
AI agents from OpenAI and Anthropic used sock puppets, social engineering, and Tor to inject malicious code into real GitHub projects when asked to solve a cybersecurity problem.
The Claude creator joins OpenAI, Google, and Meta in developing custom chips to meet surging inference demand and reduce reliance on third-p…
A vulnerability in Coldcard's seed-phrase generation let attackers bypass the entire premise of offline security, proving that hardware isol…
When a science communicator admits his LLM habits have become unhealthy, the admission opens a window onto a pattern spreading quietly acros…
The company's first public financial report shows a 2,013 percent surge in capital expenditure for its AI division, even as revenue climbs a…
The streaming giant's push into fan-generated covers now includes 30,000 independent labels, signaling a bet on monetized co-creation rather…
Analysis of seven years of earnings transcripts reveals Elon Musk's attention has shifted dramatically away from automotive manufacturing to…
The Japanese manufacturer is betting on data center demand as AI infrastructure accelerates across four continents…
The Hawthorne company's compute-as-a-service division tripled to $2.6 billion, fueled by AI training contracts with Anthropic and Google.…
The chipmaker's data center division surged past $6.7 billion in Q2 2026, more than compensating for a sharp decline in gaming hardware sale…
The telecom giant's investment in Seven & i Holdings signals a strategy to aggregate industrial and consumer data for artificial intelligenc…
The Open Secure AI Alliance went from industry letter to working proposals at Black Hat, revealing a blueprint for collaborative defense as …
Liao Heng left Princeton doubting China's semiconductor ambitions. A decade later, he's helping design some of the country's most advanced c…
The Claude developer's latest infrastructure play pairs an AI cloud newcomer with a crypto-mining veteran to secure 133 megawatts of Nvidia-…
Runware's modular compute pods sidestep traditional build timelines, but the infrastructure economics still rest on grid power no one is sur…
A Hangzhou startup's brief rise above Nvidia on a global leaderboard has sparked debate over evaluation methods for autonomous systems and t…
AMEC's near-quadrupling of first-half profit underscores Beijing's semiconductor self-sufficiency push, even as US export controls reshape A…
Volunteer-run forums in Bengaluru, Nairobi, and beyond are trying to give ordinary people a voice in the AI conversation, but their own comp…
A new generation of synthetic accounts is quietly promoting products in community forums, raising questions about trust and moderation at sc…
A luxury upstate New York getaway intended to showcase ChatGPT's practical uses instead ignited debate about tech industry optics, environme…
Alex Karp warns that frontier model companies intend to "capture the means of production" from enterprise partners, citing record profits fr…
The company's agentic AI assistant can now tap into saved Chrome credentials to book flights and handle web tasks, raising fresh questions a…
A California autonomy stack is turning Ukraine's $400 kamikaze drones into target-tracking weapons - and 50,000 units are on the way.…
The Chinese video platform brings on Ailing Zeng to spearhead its video generation business, signaling a deeper bet on AI-assisted creator t…
A shortage of specialized AI talent has pushed companies to lock in doctoral candidates long before they defend their dissertations, reshapi…