This week is still in progress. The digest is updated twice a day.
Mistral has released Timestral, a family of time-series foundation models ranging from 5 million to 209 million parameters. The largest leads the GIFT-Eval, TIME and fev-bench benchmarks among zero-shot models and runs on a CPU.
NVIDIA has published its entire Run:ai GPU orchestration platform as open source under the Apache 2.0 license. The release includes the KAI Scheduler, GPU fractioning with CUDA 13.4, quota management and a local development environment based on Kind.
The new OpenClaw release makes its update command finish and recover on its own across devices, file systems and platforms, instead of leaving a broken install. Interrupted turns, subagent work and deferred tasks now resume after a Gateway restart, so conversations are no longer lost.
Version 11.0.9.565 of the NVIDIA App adds an option that allows AI agents to control the software, and it is switched on by default. Users who have no use for agents on their machines can turn it off.
China's economic growth picked up as exports linked to AI helped make up for weak spending at home. Trade tensions and doubts about AI still weigh on the outlook.
The International Confederation of Music Publishers, which represents Sony, Warner and Universal among others, says AI has already been trained on works by Australian artists such as Kylie Minogue and Nick Cave without payment. It warns Canberra against yielding to tech companies pushing to weaken copyright law, saying Australia appears to be the only one of dozens of countries approached that is considering it.
With deepfakes now convincing enough to pass for real people on live video, World is pitching banks cryptographic proof that the person or agent on the other side of an interaction is genuine. The company launched World ID for banking and finance at the Money20/20 conference.
Since mid-October, widely followed accounts have been mixing genuine protest photos with pictures generated or altered by AI, Clubic reports. A girl holding an error-ridden sign never marched, and a Pink Panther facing a riot police van is synthetic. Elected officials and a police union have shared some of these images without checking them.
Anthropic has decided to cut off users who are too cruel to its Claude model. ZDNet France argues that the ban on insulting the chatbot is driven more by financial considerations than by moral ones.