Sakhanda Wire
NVDA $150.96 -2.89% MSFT $508.29 +0.40% GOOGL $282.44 +0.80% META $616.38 -0.40% AMZN $267.74 +3.82%

Latest news

Older stories

MarkTechPost

Code-as-World Converts Real Video into Editable MuJoCo Simulations

MirroS has introduced Code-as-World, a system that reconstructs executable representations of physical scenes from real video. Instead of treating pixels as the complete description of a world, it generates a scene.json file that MuJoCo can run, compare with the source footage, edit and simulate again. The representation records objects, geometry, dimensions, mass, friction and gravity, along with motion, contacts, camera and appearance settings. Its animation and physics engines can be used interchangeably.

MarkTechPost

Tutorial builds a custom batched ensemble with Earth2Studio

A tutorial demonstrates how to create an ensemble weather-forecasting workflow with NVIDIA Earth2Studio while preserving Colab’s existing CUDA-enabled PyTorch and NumPy setup. It loads the FCN prognostic model, obtains atmospheric initial conditions from GFS, and configures an eight-member ensemble processed in batches of two. The example saves fields including t2m, z500, u10m, v10m and tcwv, with New Delhi used as a point of interest.

TechCrunch

Sony Music and Warner Chappell sue Anthropic over alleged piracy

Sony Music Publishing, Warner Chappell and other music publishers have filed a lawsuit against Anthropic and its co-founders, Dario Amodei and Benjamin Mann. The case was submitted late Friday to the U.S. District Court for the Northern District of California. The publishers allege that Anthropic ran an illegal campaign involving torrenting, scraping and downloading copyrighted material, then used thousands of works to train its Claude AI model.

The Verge

Sony Music Publishing and Warner Chappell sue Anthropic over copyrighted works

Sony Music Publishing and Warner Chappell have filed a lawsuit against Anthropic in the US District Court for the Northern District of California. The companies seek damages related to “tens of thousands” of copyrighted works, requesting as much as $150,000 for each work and up to $25,000 for every instance in which identifiable copyright information was removed. If the maximum amounts were awarded, the claim could reach several billion dollars.

TechCrunch

Vijay Pande backs concentrated AI and biotech investments

Vijay Pande, who previously led a16z’s biotech practice managing nearly $4 billion, has shifted to a smaller investment firm, VZVC. Co-founded with Zach Werner, the firm plans to make about five highly concentrated investments a year rather than 30. It has no associates and uses AI agents for much of its operations.

MarkTechPost

Google updates Gemini Omni Flash with longer scene extensions and 4K output

Google has released Gemini Omni 1.1 Flash, an update to its multimodal video generation and editing model. The model can use up to 10 seconds of preceding video when extending a scene, compared with the final second in earlier versions. Extensions are made in 3–10-second increments and can reach 40 seconds in total. The feature adds footage only to the end of a clip, while uploaded source videos must generally be no longer than 10 seconds.

The Decoder

OpenAI to end Cursor contract after SpaceX acquisition

OpenAI plans to terminate its contract with the AI coding tool Cursor on November 12, 2026, following SpaceX’s acquisition of the company. OpenAI said the agreement allows termination within a limited period after a change in ownership and that November 12 is the longest notice period permitted. The company said it could no longer be confident that SpaceX would follow its terms of service. Thibault Sottiaux described the decision as a matter of trust.

The Decoder

AI videos replace human performers in China’s entertainment sector

AI-generated video is increasingly being used instead of human performers in China’s entertainment industry, affecting actors, influencers and livestreamers. After ByteDance launched Seedance 2.0, digital performers have been able to create videos faster and at lower cost than people, according to the report.

TechCrunch

Nvidia’s advantage expands from GPUs to data-center orchestration

Nvidia’s position in artificial intelligence is increasingly tied to the systems surrounding its GPUs, not only to the processors themselves. As AI computing expands to gigawatt scale, operating very large data centers efficiently requires increasingly complex coordination of memory, storage, networking and computation. The shift follows growing competition from hyperscalers such as Amazon and Google, which have developed their own chips.

The Verge

AI-generated artists gain chart placements and major recording deals

Fully AI-generated songs and virtual performers are beginning to achieve measurable commercial success. Last summer, Vinih Pray’s Suno-created track “A Million Colors” reached No. 44 on TikTok’s Viral 50 chart. AI artists Breaking Rust and Cain Walker also appeared on Billboard’s Country Digital Song sales chart, which tracks individual song downloads.

The Decoder

Open video dataset offers 10 million hours for AI research

LAION has released the Big Video Dataset (BVD), an open resource intended for AI research. It was built from 1.3 billion video URLs identified in CommonCrawl. The team downloaded 80 million videos, representing 10 million hours of material, and created 55 million clips with automatically generated descriptions of their video and audio. The release also includes 300 million still images.

The Decoder

Anthropic proposes a common standard for controlling physical devices

Anthropic is developing the Model Hardware Standard (MHS), a specification intended to give AI agents a common way to read data from and control programmable devices such as microscopes, robotic arms and laboratory instruments. The project was developed with HHMI Janelia Research Campus and is being offered initially as a research preview to selected laboratories and manufacturers; an open-source release is planned later.