AIWadmin
Microsoft builds a benchmark that checks what agents actually change
A new Microsoft and Hugging Face benchmark grades agents on the records they leave behind, not the reports…
A Harvard physicist hands Claude exact math and finds new answers
An open-source harness built around Claude solved a two-decade-old ecology equation and produced fresh results in particle physics.
DeepMind essay says intelligence grows from teamwork, not a singularity
A Google-affiliated essay argues advanced AI will arrive as a social system of people and agents rather than…
Audit finds Chinese AI models push Beijing’s talking points
A new benchmark says leading Chinese models repeat official positions or refuse to engage on sensitive political questions.
Google method stops self-improving agents from acing their own tests
A new training approach keeps agents from memorizing their practice tests while trimming the compute they burn.
Google cuts free Gemini access back to one small model
Google is pulling its stronger Gemini models behind a paywall and leaving free personal accounts with the smallest…
IBM lets banks run its coding agent inside their own walls
IBM made a self-hosted version of its Bob coding agent generally available, letting enterprises run agentic software development…
Suno opens a Speech beta that narrates and scores at once
Suno opened a Speech beta that generates narration and its background music together, as the company pushes beyond…
PwC finds no one really owns AI risk at most companies
A PwC survey of 4,000 business and tech leaders found no clear owner for agentic AI security, with…
Amazon commits $1B more to towns hosting its data centers
AWS will add more than $1B over five years for schools, energy upgrades and local grants in US…
Meta hands builders the firmware to wire Muse into any gadget
Meta open-sourced ESP32 and Linux SDKs that let tinkerers connect its Muse assistant to screens, sensors and smart-home…
Sean Parker steers Stability AI toward paid work for musicians
Napster co-founder Sean Parker is rebuilding Stability AI into a music-production toolmaker, two years after helping rescue the…
Microsoft’s streaming speech model transcribes while you talk
Microsoft released MAI-Transcribe-2-Streaming, a real-time speech model it says tops Artificial Analysis accuracy while returning draft text in…
Nvidia adds a cheaper 64GB Spark desktop for always-on agents
Nvidia's new 64GB DGX Spark desktop, shipping October 23 from $4,999, runs 30B-class open models locally and clusters…
PaleBlueDot AI banks $200M to expand its GPU cloud for frontier labs
Palo Alto-based PaleBlueDot AI raised a $200M Series C led by ComputeCore at a $3.2B valuation to add…