The US weighs restrictions on open-weight AI models as Nvidia and Mistral push back, highlighting a growing industry divide over...

AI safety guardrails that prevent models from writing exploits are slowing down legitimate security researchers who need them for defense.

AMD unveils Helios, a full rack-scale AI system taking direct aim at Nvidia, signaling it wants the entire data center,...

Claude Voice Mode gets smarter models from Anthropic, enabling the assistant to handle calendar rescheduling and complex voice tasks.

Etched hits $10.3B valuation building GPU-free AI chips, convincing top investors it can challenge Nvidia without traditional GPU architecture.

ServiceNow makes a $40M bet on BusinessNext, a specialized banking AI platform most people outside Indian banking have never heard...

DeepSeek open-sourced DSpark, a speculative decoding framework it says makes V4 up to 85% faster, no retraining or new hardware...

DeepSeek's DSpark speeds up its V4 models by as much as 85 percent per user without new hardware. The code...

IBM's chief executive pushed back on the obituary directly, arguing that AI has not made the mainframe obsolete.

OpenAI believed it had built a highly isolated sandbox, but experts who examined what happened concluded the isolation had a...