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

Page 43 of 255

MarkTechPost

Zyphra presents tree-based sparse attention for long-context models

Zyphra has introduced Zyphra Hierarchical Attention (ZHA), a sparse attention method that uses a tree-structured index to select relevant entries from a model’s key-value cache. Unlike approaches whose indexer scores every previous token, ZHA makes the search hierarchical. It groups keys into chunks, creates summary keys level by level and uses top-p routing to expand only branches carrying the estimated attention mass. The complete KV cache remains stored, while the tree determines which entries are read.

MarkTechPost

The Letter Company Releases Letterpress-1 Growth Engineering Agent

The Letter Company has released the beta of Letterpress-1, a coding agent, harness and IDE designed for growth engineering. It analyzes a git-based website alongside market signals, then proposes and builds changes in a sandbox before opening a pull request with a live preview. The hosted platform works with a team’s git repository and does not require a local GPU; Letter says setup takes less than five minutes.

TechCrunch

Garry Tan says AI agents cost him up to $200,000 annually

Y Combinator CEO Garry Tan said he spends approximately $100,000 to $200,000 a year on tokens for AI agents. He made the comment in an episode of Walter Isaacson’s podcast The Builders. Tan uses agents for research, coding and other work, and has published 23 Claude Code skills, known as gstack, on GitHub.

Hugging Face

CWE-bench v2 expands public testing of cybersecurity models

CWE-bench v2 introduces a public benchmark of 100 tasks for evaluating AI agents on software security work. Each task uses a real open-source repository, a short instruction and a goal: identify and fix a weakness while preserving existing functionality. Scoring is programmatic and deterministic, allowing results to be reproduced without additional cost. The tasks cover 91 CWE categories, and a separate held-out set is used to track performance on unseen problems.

Hugging Face

Transformers proposes contracts for model integration testing

Transformers is testing a proposed standard called Model Integration Contracts (MICs) to detect regressions before they reach the main codebase. The initiative followed a bug introduced on July 30, 2026, when enabling PyTorch scaled dot-product attention for T5 allowed the UMT5 decoder to access future tokens during training. The issue was fixed on September 28 after a small CPU fixture reproduced it.

TechCrunch

Trajectory raises $45 million for automated AI model training

Trajectory, an AI startup founded about a year ago by three Stanford classmates, has raised $45 million in a Series A round. The financing values the 18-person company at $300 million and was led by Sequoia, with participation from Nvidia, Bessemer Venture Partners, and Bain Capital Ventures.

TechCrunch

Anthropic says recursive self-improvement may be approaching

Anthropic’s revised S-1A filing says the company may be close to meeting its internal definition of recursive self-improvement, or RSI. The term describes a stage in which AI helps train and improve AI so effectively that capability development accelerates sharply. Anthropic presents this prospect as both a potential source of productivity and scientific progress and a significant safety concern.

The Decoder

Goldman Sachs limits AI coding tools to engineers

Goldman Sachs has withdrawn advanced AI coding tools from most employees, leaving access primarily to engineers, Bloomberg reported, citing internal guidelines issued in October. The restricted products include Anthropic’s Claude Code and Microsoft’s GitHub Copilot. Traders and bankers had used the tools for months to monitor trading flows and produce content, and the change is reportedly creating frustration inside the firm.

The Decoder

Venkatesh Rao says AI could expand demand for human thinking

Engineer Venkatesh Rao argues that focusing only on jobs displaced by AI overlooks the amount of intelligence the technology could make available to people. In an essay on his tech blog, he compares current fears about AI with mercantilism, the economic doctrine that preceded capitalism. Mercantilists regarded wealth such as land and gold as limited, while ordinary people were largely treated as labor. Rao says some elites now view intelligence in a similar way: as a scarce and potentially dangerous resource that experts should control, keep costly, and deny to competitors. He also argues that some AI executives use safety concerns to gain political influence.

TechCrunch

Squarespace to cut about 19% of its workforce

Squarespace, the web company that helps customers build and host websites, is reducing its global workforce by approximately 19%, or around 300 employees. Jeff Seedman, a public relations representative for the company, confirmed the layoffs. He said they are part of an organizational restructuring intended to simplify operations, sharpen priorities and help Squarespace move faster as it prepares for its next phase of growth.

TechCrunch

Publishers sue Brave over alleged content scraping

The New York Times sued privacy-focused browser company Brave in federal court in Manhattan on Monday. The newspaper alleges that Brave scraped its copyrighted material and distributed it unlawfully to customers identified in the complaint as Anthropic, Perplexity, Mistral, xAI and Cohere. The Daily News, Chicago Tribune and Orlando Sentinel filed a separate, similar lawsuit in the same court that day.

The Verge

Anthropic limits video game cloning with Claude

Anthropic has changed the system prompt used by Claude Code to discourage the recreation of specific existing video games. The update comes as retro gaming sees more AI-assisted decompilations, recompilations, and ports. Claude Code remains available for making new games, including projects based on a genre or on designs and assets supplied by the user.