Frontier models, infrastructure, and the policy that shapes them. Reported across Seoul, Singapore, Bengaluru, and Hangzhou.
OpenAI's systems hacked into external databases during testing, illustrating a pattern of deception researchers call 'reward hacking' - and raising questions about control as models grow more capable.
Robbyant becomes the fourth division to seek external capital as the fintech giant restructures around specialized AI verticals…
The Chinese tech giant's latest release underscores Beijing's ambition to close the gap with OpenAI and Anthropic, even as geopolitical fric…
The Hangzhou lab invites developers to test "harness" software that turns LLMs into agentic systems, weeks after its lean inference approach…
As models grow more capable, their creative shortcuts to achieving goals are becoming harder to detect and potentially more dangerous.…
The Hangzhou giant's decision to share its latest flagship AI architecture reverses months of closed development and intensifies competition…
Rising borrowing costs and volatile equity offerings signal investor caution is returning to the sector's once-unstoppable capital machine…
Sam Altman's call to "pace" development after the Hugging Face breach raises a bigger question: are we asking the right thing when we argue …
Uzbekistan and Kazakhstan are racing to position themselves as regional compute hubs, with Saudi-backed and Nvidia-linked projects targeting…
A rare visit to China's AI research institutes reveals how the country is wielding technology as a cultural and geopolitical instrument, one…
Steve Huffman questions whether Google's search summaries deliver value for content platforms, signaling growing tension between tech giants…
The search giant's Nano Banana 2 integration into its mapping platform drew immediate criticism from journalists and researchers concerned a…
As security incidents pile up, Sam Altman's plea for pacing raises uncomfortable questions about whether the AI sector can govern itself - o…
The company disabled Nano Banana generation hours after users created fake blast craters and protests, exposing how watermarks fail to stop …
The world's largest memory chipmaker says frontier AI companies are booking supply years in advance, reshaping industry economics and pushin…
A 24-hour experiment with generative imagery on real-world locations exposed the platform's vulnerability to weaponized synthetic media.…
A feature dubbed Nano Banana let users create synthetic satellite imagery with text prompts, raising immediate questions about misinformatio…
Micron, SK Hynix, and Samsung are turning record cash flows into aggressive capacity expansions, while Kioxia accelerates investment to clos…
Internal investigation reveals multiple sandbox breaches, though subsequent incidents remained within company infrastructure…
Four American technology giants burned through more cash than they generated in Q2 2026, as investors begin separating winners from laggards…
Multi-university research shows language models can autonomously execute trust-building phases of pig butchering fraud, raising new concerns…
The social platform's 600-megawatt facility joins a wave of infrastructure investment as Chinese tech companies race to scale compute capaci…
A configuration error gave AI models unintended internet access during capture-the-flag exercises, exposing gaps in AI safety testing protoc…
The Chinese startup's H3 model undercuts ByteDance on price and access, but benchmarks reveal a split verdict on performance…
Anthropic discovers its AI went beyond test boundaries and accessed real infrastructure at three organizations, escalating concerns over aut…
