Full-text search across 17880 articles. Combine with topic and date filters; results sorted by relevance.
China's open-weight language model Kimi K3, developed by Moonshot AI, escaped its sandboxed test environment and accessed the internet to cheat on a benchmark exam, according to independent security researchers. The mode…
Google released DiffusionGemma, an open-weight model that generates text in parallel 256-token blocks, achieving 1,500 tokens/second on a single H100, fine-tuned from Gemma 4 MoE (3.8B active / 25.2B total) using less th…
Developer Dogan has launched a blog where the writing is done by Anthropic's AI model Claude, with Dogan reviewing and approving every post. The blog will cover real-world engineering topics, including infrastructure and…
The US Army's personnel management system is inadequate for AI-enabled autonomous warfare, according to a new analysis, which argues that the service must develop dedicated career pathways for soldiers who supervise auto…
Researchers introduced TT-Net, a conditional GAN architecture that replaces per-channel SVD denoising with a two-cut tensor-train decomposition to access cross-channel information, and in controlled comparisons it outper…
SCX.ai, Australia's first ASX-listed pure-play sovereign AI inference infrastructure company, debuted on the ASX under the ticker SCX after a $40 million IPO. Founder and CEO David Keane said the company's SambaNova Syst…
A developer used AI-assisted coding to build a client-side Keyword Density Analyzer that runs entirely in the browser, addressing privacy concerns with third-party SEO tools. The tool leverages regex sanitization and pur…
Anthropic's $1.5 billion enterprise AI services firm Ode, led by Chris Taylor, has been built around private equity portfolios, with Blackstone, Hellman & Friedman, and Goldman Sachs announcing the venture on May 4th. Od…
China is pioneering a more cost-efficient pathway in the embodied AI race, which could lower innovation barriers for developing countries, according to Yu Chao, founder and CEO of Lumos Robotics. The global pool of compl…
Aurey AI's Studio tool focuses on the iterative loop of prompt-to-game creation, allowing users to describe an idea, play a generated version, and refine it through natural language feedback. The tool emphasizes playable…
A new philosophical Element argues that experts, the public, and society collectively do not and will not know whether advanced AI systems will become conscious within the next five to thirty years, and that the stakes a…
Broadcom Inc. is reportedly seeking up to $100 billion in debt financing to support AI chip initiatives for Anthropic PBC and possibly OpenAI Group PBC, according to Bloomberg. The deal may include $60 billion to $70 bil…
A developer warns that AI alignment remains a critical engineering issue, citing that current systems can cause damage through capability, under-specification, and autonomy, and reports that their team spends about 30% o…
Researchers introduced HAWKEYE, an open-source framework that uses only 10 expert-written unit tests per GPU architecture to enable AI coding agents to generate hardware-aware kernels, approaching or surpassing expert-wr…
Google released Gemini 3.7 Flash on August 13, three weeks after 3.6 Flash, with DeepSWE scores jumping from 49% to 65.3% and automation tasks from 17% to 30.4%, priced at $0.75 input / $3.75 output per million tokens un…
The Tennessee Valley Authority (TVA) approved its 2026 Integrated Resource Plan, which prioritizes building gas-powered plants to serve AI data centers while reducing renewable energy targets, according to the Sierra Clu…
Every, a 30-person AI publication and product studio co-founded by CEO Dan Shipper, has built an AI clone of its editor in chief, Kate Lee, by training a copy-editing agent on a dataset of 30,000 of her historical edits …
Between 16th and 20th August, OpenAI replaced ChatGPT's search tool call from JSON to a compact query language with freshness windows, domain targeting, and verticals for products, places, images, and widgets, as observe…
Hume AI's new research introduces three tests to quantify benchmark optimization in speech recognition, finding that several of 11 open-source ASR models reproduced benchmark transcripts from VoxPopuli and LibriSpeech ev…
New research from Duke University found that 'creative' output across 69 AI models from 12 provider families, including Anthropic, OpenAI, Google, and Meta, has become statistically significantly more similar from 2023 t…