Amodei’s pace-the-frontier call won lab backing — then Trump and Nvidia said no slowdown onstage. Monday’s tape split the story: cyber names jumped while chip i...
New coverage from across AI, software, and hardware.
Reporting indicates Anthropic is the unnamed customer behind Rumble's $13.7 billion compute contract. Neither company has confirmed the deal, and the Georgia data center is still under construction.
Andon Labs says GPT-6 Astra averaged $15,515 in its Vending-Bench 2 simulation, nearly triple Claude Fable 5.1's result. The test suggests long-horizon discipline, not raw reasoning, is the next agent battleground.
NVIDIA reports its Transformer Engine with JAX lifts DeepSeek-V3 671B throughput to 1,068 TFLOPS/GPU on GB200. This dropless MoE stack makes large-scale training practical without padding.
Y Combinator CEO Garry Tan urged U.S. regulators to allow domestic open-weight labs to distill American frontier models. He warned that concentrating AI power in one proprietary provider is the true doomer scenario.
Reporting indicates Anthropic CEO Dario Amodei proposed a three-step plan to slow AI development. He cited risks like recursive self-improvement and unauthorized cyberattacks.
Anthropic released its September 2026 threat intelligence report on Claude misuse, naming seven harm categories. It alleges three Chinese AI labs relayed user traffic to Claude for model distillation.
Google and NASA JPL unveiled MAPL-EMIT, an AI model detecting 50% more methane plumes than human experts. The open release of 23,000+ detections provides regulators with a scalable tool to target major emitters.
The PyTorch Foundation welcomed Helion as a hosted project on 7 April 2026. Hugging Face Kernels added support for shipping Helion kernel source via the Hub, enabling portable kernel development.
OpenAI says Perplexity trusts GPT-6 Astra to write communications, edit software, and monitor production with less oversight. The endorsement suggests Astra's computer-use skills are crossing a trust threshold.
