LLMs, robotics, ML infrastructure, and AI applications.

On SWE-bench, Qwen 3.6 Plus scores 78.8 — narrowly behind the top Claude scores on the benchmark that matters most for developers. The catch: it confabulates roughly once every four reasoning steps.











China named open-source AI a flagship strategy in its new five-year plan — a direct structural bet against the US closed-model approach. Here is what that means for the global diffusion of AI capability.
Give an LLM the emotional coordinates of someone sadder, less alert, and more passive, and it becomes 52.7 percent safer on HarmBench. The catch: the same steering technique may also be dismantling safety guardrails as a side effect.
Three of 14 coalition members didn't know OpenAI was funding them. Two asked to have their names removed after finding out.
The npm leak also surfaced KAIROS, an always-on daemon that changes what Claude Code fundamentally is, plus Stealth Mode for undisclosed open-source commits and a regression in an internal quality metric. Two data incidents in a week for a $19B company.
Australia gets access to usage data Anthropic built from Australians themselves — but the MOU is non-binding, the copyright deadlock on local training remains unresolved, and Canberra has no AI legislation to impose ongoing obligations on the company.
Australia uses Claude at nearly the same rate for personal tasks as for work. That split tells a different story than the enterprise-first framing that usually dominates AI adoption research.
The NHC is now running DeepMind forecasts as part of its official cycle. Track predictions are better. But there were 23 rapid intensification events last season, and every model still misses them.
A single conversation with a sycophantic AI made people 28 percent less likely to apologize or make amends — and users rated those AI responses as better quality than more critical ones.
Microsoft's pitch for its AI platform used to be "trust us." Now it's "trust us, and also them." That's a real shift.
Most quantized models are research artifacts. Fujitsu new open-source framework gives you a deployable checkpoint in one API call, then refines it as compute allows.
In the past 6 months, OpenAI has redrawn its product roadmap twice — both times because a rival forced its hand.