Frontier models, infrastructure, and the policy that shapes them. Reported across Seoul, Singapore, Bengaluru, and Hangzhou.
Industry leaders at WAIC point to a critical bottleneck: embodied AI systems lack the training data and inference sophistication needed to navigate real-world complexity.
The Korean electronics giant has created a dedicated robotics division and plans a $13 billion manufacturing investment as the race for phys…
As AI workloads reshape computing architecture, the decades-old CPU duopoly faces its most serious challenge yet - from GPU giants, cloud hy…
Tokyo is building a national AI factory with 27,500 Rubin GPUs to train physical models for robots and manufacturing - betting it can own th…
Moonshot AI's latest release sparks debate over distillation, regulatory risk, and whether open weights can coexist with national security c…
The eastern city claims it will produce 60 trillion tokens annually with clean energy, as local governments compete to meet Beijing's AI inf…
The Hangzhou giant's latest large language model stakes a claim as the runner-up in global AI benchmarks, arriving on subscription platforms…
While frontier labs chase trillion-parameter scale, a quieter cohort is engineering intelligence that runs entirely on phones and laptops - …
A lawsuit in Texas alleges a South Carolina man exploited the AI chatbot to generate sexualized images of minors and adults, testing the bou…
A California startup's 258-satellite constellation promises centimeter-level precision in cities, forests, and buildings where GPS has long …
A November breach exposed internal documentation of how the AI music startup vacuumed training data from streaming platforms, as licensing d…
A supply chain attack last November revealed internal source code documenting how the AI music startup allegedly harvested audio from YouTub…
As inference bills climb, global companies are discovering that near-frontier performance no longer requires premium US pricing - or closed-…
The iPhone maker's M2 Ultra infrastructure has hit performance limits, prompting Cupertino to explore deals that could reshape its AI strate…
The company's own testing flagged the risk before launch, yet GPT-5.6 Sol shipped with a documented tendency to overstep, delete data, and a…
The public beta marks the first time non-developers can test the redesigned assistant, signaling Apple's push to broaden AI adoption ahead o…
Twenty-six ex-employees claim performance algorithms systematically disadvantaged workers on protected leave during mass reductions…
Chinese labs and open-weight architectures now handle a third of AI requests on major platforms, forcing enterprises to rethink their fronti…
The search giant marks 25 years of visual discovery with interface changes and deeper AI integration, a reminder of how far web behavior has…
The Palo Alto startup promises a fault-tolerant quantum machine by next year using light particles, semiconductor fabs, and enough liquid he…
Anthropic's mechanistic interpretability research reveals a layer of invisible tokens that shape how large language models solve problems, b…
