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

MarkTechPost

Google Open-Sources TorchTPU for Native PyTorch on Cloud TPUs

Google has released TorchTPU, an open-source PyTorch backend for Google Cloud TPUs developed with Meta’s PyTorch team. The project was announced at PyTorch Conference 2026 on October 20, 2026. Existing PyTorch programs can select a TPU through torch.device("tpu"), placing TPUs alongside CPUs and GPUs as native devices. Google recommends TorchTPU for new workloads as a replacement for PyTorch/XLA.

MIT Tech Review

Enterprise AI economics shifts toward utilization and business outcomes

Enterprise AI is moving from isolated chatbot and copilot trials to a production layer connecting models, agents, data, and business workflows. The article argues that token or GPU-hour prices alone do not show the full economics. A more useful measure is the effective cost of a business-relevant result, including platform, model, operations, and capacity expenses.

Hugging Face

Learned memory lifts GPT-4.1 to the top of τ²-bench retail

ALTK-Evolve uses GPT-4.1 to extract behavioral guidelines from an agent’s previous task attempts and adds them to the system prompt. The model weights are not changed, and no human annotation is used. On τ²-bench retail, the resulting ReAct agent achieved a 67.5% Pass^4 score, meaning it completed all four trials for that share of tasks and exceeded every retail entry in the October 12, 2026 leaderboard snapshot.

MarkTechPost

DeepSeek open-sources the DeepInfer inference framework

DeepSeek has released DeepInfer, the inference framework used to serve DeepSeek-V4.1-Flash in production. The project is available under the Apache 2.0 license, while the company’s deepseek-recipe token encoding and decoding layer uses the MIT license. DeepInfer is designed for offline inference and supports one model with a 1M-token context window.

The Verge

Spotify removes over 50 million AI and spam tracks from recommendations

Spotify says it has removed more than 50 million tracks classified as AI-generated slop or spam from users’ recommendations during the past year. The company also adjusted its recommendation system to give greater prominence to music created by human artists. Spotify says its editors do not program content identified as slop or spam, and that listener concerns about such material appearing in recommendations have declined since the changes.

TechCrunch

Figma expands MCP support despite co-founder’s protocol concerns

Figma is broadening its support for the open-source Model Context Protocol (MCP), which defines how AI agents connect to external data and tools. At the company’s first Config conference in India, it announced 20 new MCP partners, including Amp, ClickUp, Dia, Junie by JetBrains, Linear, Meta Muse, Mistral, Notion, OpenCode and Sarvam Agents.

TechCrunch

U.S. orders civilian agencies to remove DeepSeek from government systems

The United States has directed federal civilian agencies to stop buying products and services developed by Chinese AI startup DeepSeek and to remove them from government IT systems. The Department of Homeland Security said the measure is linked to national security concerns about the company’s operations in China.

TechCrunch

Eka Robotics raises $150 million for physical AI manipulation system

Eka Robotics has raised a $150 million Series A round at a post-money valuation of $1.45 billion. The financing was led by a16z and F-Prime. The startup was founded by MIT professor Pulkit Agrawal and former Google DeepMind researcher Tuomas Haarnoja, and is developing a robotic manipulation system built around a force-sensitive gripper and a deep-learning model.

Ars Technica

Meta pauses Muse conversation sharing over privacy review

Meta is suspending Muse’s conversation-sharing feature after researchers reported that the AI agent’s public links could expose information users did not intend to make available. The company said it is reviewing the behavior and considering changes to the feature.

The Decoder

Cloudflare introduces Clairic for faster document data extraction

Cloudflare has introduced Clairic, an AI model designed to extract specified information from unstructured documents. It is available as a preview through Workers AI, while Cloudflare says open weights will be released later. Users identify fields such as an order number or invoice total, and the model returns values in the requested format together with a confidence score. Missing information is represented as “null.”

MIT Tech Review

AI adoption is spreading from technology into other industries

A 2026 TripleTen survey indicates that artificial intelligence is being introduced at different speeds across industries, with technology companies at the forefront and finance following. In finance, 48% of workers said they had been asked to use AI chatbots, while 32% had been asked to automate part of their work. The corresponding figures were 18% for chatbot use in healthcare and 25% in education. The article attributes the slower pace in these sectors partly to specialized, public-facing and compliance-heavy work.