Moonshot releases Kimi K3: a 2.8-trillion-parameter open-weight model, 1.56 terabytes on disk
Moonshot AI published the full downloadable weights for Kimi K3, a 2.8-trillion-parameter model that uses only 104 billion parameters per token, handles text and images, and reads just over a million tokens of context.
Kimi K3 is downloadable, but the floor to run it is eight datacenter GPUs
Kimi K3's 1.56-terabyte checkpoint needs a single eight-GPU B300 or MI355X node as its practical minimum, and no version of llama.cpp can load it today, so open weights currently mean operator-scale rather than local.
Anthropic says it never asked to ban open-weight models, and names what it does want instead
Anthropic published its position on open weights, rejecting a categorical ban while backing three specific restrictions: chip export controls, action against industrial-scale distillation, and mandatory pre-release safety testing for sufficiently capable models, open or closed.
NVIDIA launches an open AI security alliance with 41 partners, and OpenAI is not on the list
NVIDIA announced the Open Secure AI Alliance with 41 inaugural partners including Microsoft, the Linux Foundation, Hugging Face and CrowdStrike, built around the claim that closed APIs blocked forensic work during the Hugging Face breach while an open model did it.
Sysdig documents JadePuffer, an AI agent that ran a database extortion attack end to end
Security firm Sysdig documented an intrusion in which an AI agent chained a known Langflow flaw into a full database extortion attack without a human approving each step, encrypting 1,342 configuration records and fixing its own failed login in 31 seconds.
Adding skills to an AI agent breaks work it already did right, and the cost cancels most of the gain
A new study measured what installing skill libraries does to office-work agents and found they newly solved 553 task conditions while breaking 324 that the plain agent had already handled, cancelling 59% of the gross improvement.
Starbucks pulls its AI inventory counter from 11,300 cafes after nine months
Starbucks has retired Automated Counting, an iPad-based computer vision system for counting shelf stock, nine months after rolling it out to 11,300 company-operated cafes, after workers reported miscounts and lost progress.
OpenAI says one in six work prompts is a task from someone else's job
OpenAI analyzed more than 800,000 work-related ChatGPT messages and found about one in six were classified as tasks belonging to a different occupation than the user's own, rising to nearly half once generic work is excluded.
NVIDIA invests in Safe Superintelligence, promising an order-of-magnitude compute jump
NVIDIA and Ilya Sutskever's Safe Superintelligence announced a long-term strategic partnership including an NVIDIA equity investment, which SSI says will expand its available compute by an order of magnitude.
Terence Tao says the bottleneck in AI-assisted mathematics is understanding, not proofs
In an ICM public lecture, Terence Tao argues that AI and formal proof systems accelerate generating and verifying proofs but not explaining, reviewing or canonicalizing them, so correct results could pile up faster than the field can absorb them.
Leaked Suno code names YouTube, Deezer, Genius and podcast RSS feeds as collection sources
A hack of AI music generator Suno exposed source-code files and comments naming YouTube Music, Deezer, Genius, stock libraries and podcast RSS feeds as data collection targets, the most specific provenance evidence yet in the music industry's copyright fight.