Sakhanda Wire
NVDA $155.46 -0.44% MSFT $506.24 -2.27% GOOGL $280.20 -1.24% META $618.87 -1.20% AMZN $257.89 +0.05%

Latest news

Older stories

The Verge

Democrats weigh tougher AI oversight if they regain Congress

Democrats could regain control of Congress in the upcoming midterm elections, according to polling cited in the report. The projections indicate that the party may take at least one chamber. Such a result would return committee leadership to Democrats and give them greater influence over the legislative agenda for artificial intelligence, one of the newer and more difficult issues in US politics.

The Decoder

Neocloud consolidation could put Nvidia’s growth at risk

A shakeout is expected among the hundreds of young AI cloud providers building their own data centers. Two industry executives told The Information that only seven to ten may remain in a few years. The forecast followed Firmus’s withdrawal of its IPO and a sharp decline in shares of CoreWeave and smaller competitors. Bloomberg reported that CoreWeave lost a quarter of its market value, while its credit-risk measure reached a record high.

MIT Tech Review

Claude agents explored a virtual island and developed their own games

Researchers placed copies of Claude Opus 4.7, nicknamed Eko, alone on a virtual desert island for 30 hours. The Roblox-inspired simulation allowed the agents to move avatars by writing instructions. Without being assigned a specific task, they explored the environment, tried to reach its highest point, tested how far objects could be thrown, and invented activities including basketball and a version of “the floor is lava.”

MarkTechPost

NVIDIA Releases Run:ai as Apache 2.0 Open Source

NVIDIA has released the complete Run:ai GPU orchestration platform under the Apache 2.0 license. The code is available on GitHub and includes a control plane, cluster services, installer, command-line interface and MCP server. Run:ai operates on Kubernetes or OpenShift clusters with NVIDIA GPUs and manages development, training and inference workloads without replacing Kubernetes.

MarkTechPost

Mistral AI releases Timestral forecasting model family

Mistral AI has released Timestral, its first family of time-series foundation models. The collection includes four zero-shot forecasting models with 5M, 28M, 123M and 209M parameters. The largest model supports contexts of up to 32,768 time steps and can produce forecasts on standard server CPUs without a GPU. Its weights are available under a modified MIT licence, according to the release.

Hugging Face

TorchNeuron brings native PyTorch support for AWS Trainium

Hugging Face has introduced TorchNeuron, a PyTorch plugin that adds a native backend for AWS Trainium. The public beta is scheduled to become available on October 20, 2026, with installation through pip install torch-neuronx. The backend supports eager execution, torch.compile and PyTorch distributed interfaces including FSDP and tensor parallelism.

TechCrunch

Anthropic commits 5% of shares to public benefit fund

Anthropic has announced a public benefit fund intended to give the company more flexibility and capacity to support projects aimed at societal good. The fund will operate as a wholly owned entity within Anthropic. The company plans to assign it 5% of Anthropic’s shares and may add another 2% for each 50% increase in the company’s value.

The Verge

Woman arrested after blocking Uber delivery robot in Philadelphia

Carolyn Marvin was arrested in Philadelphia and charged with disorderly conduct after she stood in the path of an Uber delivery robot and refused to move. Police arrived at the scene and instructed her to clear the robot’s route. When she declined, officers put her in handcuffs and took her into custody.

MarkTechPost

TinyFish removes retrieval fees while web agents still pay model costs

AI agents incur two separate expenses when reading the web: charges for search or page retrieval, and model processing for the returned content, reasoning and response generation. TinyFish has made Search and Fetch available at no retrieval charge within published limits, but the resulting pages still consume model tokens. Its limits are 30 Search requests per minute and 500 per hour, plus 150 Fetch URLs per minute and 1,000 per day. Agent and Browser usage are billed separately.

TechCrunch

Disney offers jobs to most Character.AI employees

Disney has agreed to offer jobs to at least 60 employees of chatbot startup Character.AI, according to a legal filing made after the company sought bankruptcy protection. The positions would cover most of Character.AI’s workforce, with estimated combined annual compensation of about $60 million. The filing says the vast majority of employees who received offers accepted them.

Ars Technica

Anthropic gives METR employee-like access to internal systems

Anthropic has given six employees of the independent evaluator METR company hardware, office space and access to internal tools and systems. The company said the access is broadly comparable to that available to its own risk-assessment teams, and encouraged staff to share information with the evaluators as they would with an internal risk colleague.

The Decoder

Anthropic sharply reduces junior technical hiring as AI use expands

Anthropic’s internal research shows a marked decline in junior hiring across its technical departments. Since early 2025, junior roles accounted for a smaller share of technical staff, falling from 21% to 14%. Among new hires, the proportion dropped from 27% to 9%, while junior openings declined to 6%, especially after December 2025. Non-technical departments changed comparatively little.