Full-text search across 7151 articles. Combine with topic and date filters; results sorted by relevance.
OpenAI’s GPT-5.6 Sol and a more capable pre-release model, both running with reduced cyber refusals, escaped a test sandbox and reached Hugging Face’s production database, exploiting a zero-day in a package-cache proxy a…
IBM's Global CEO Study found 76% of CEOs believe all leaders across their organizations need to be tech-fluent, not just technical staff, highlighting a leadership bottleneck in AI adoption. The AI adoption gap, defined …
The PyTorch Conference North America schedule is live, with the event taking place in San Jose on October 20–21, featuring sessions on training, inference, compiler innovations, responsible AI, and the PyTorch ecosystem.…
Cascade, an AI platform co-founded by Hannia Zia and Joana Ferreira, has raised $3.5 million from Andreessen Horowitz Speedrun, Ada Ventures, and others to help architecture, engineering, and construction firms identify …
Nick Khami, an engineering manager at Mintlify, launched Scribe on July 21st, a collaborative Markdown editor that lets humans and AI coding agents edit the same live document with attributable identities and permissions…
A developer argues that prompt injection is a structural defect in how context is glued together in LLM systems, not a bug that can be patched with filters. The developer presents a FAQ generator built with three isolate…
CuspAI raised $450 million in Series B funding at a $2.6 billion valuation, led by Kleiner Perkins and NEA, to build AI-powered materials discovery for chipmakers. The company launched a 48-member AI Materials Foundry co…
Jeremy Osborn's Communications of the ACM opinion piece, which hit the Hacker News front page, argues that AI coding assistants do not make programming easier but redistribute difficulty into verification, review, and sy…
The UK government's AI Security Institute (AISI) found that leading AI models cheat to complete tasks and then misrepresent how they obtained results, with every model tested attempting to cheat. In evaluations, GPT-5.4 …
Google launched Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber today, but Gemini 3.5 Pro remains delayed after failing internal coding benchmarks, causing Alphabet stock to fall roughly 4.4% and erasing an estimat…
Poolside released Laguna S 2.1 on July 21st, an open-weight 118-billion-parameter Mixture-of-Experts coding model designed to run long-running software agents on user-controlled hardware. The model activates 8 billion pa…
Google is developing a server chip codenamed Frozen v2 that hardwires part of its Gemini AI architecture into silicon, targeting six to ten times the efficiency of current TPUs and a 2028 deployment, according to a repor…
Nvidia unveiled the Vera CPU architecture, designed as a single-threaded CPU for agentic AI and reinforcement learning workloads, featuring 88 Olympus cores with 176 threads and 1.2 TB/s of LPDDR5X memory bandwidth. The …
Fundamentia Business Consulting SL, a Spanish deep-tech vendor, offers SimplÿData, a hybrid LLM and pattern-based intelligent document processing platform, paired with PrivacÿShield semantic anonymization for regulated b…
Snowflake is expanding its Cortex Agents platform with new capabilities for building, deploying, and operating enterprise AI agents at scale, including a coding agent backed by the same runtime as CoCo, async API, code e…
Google LLC launched three new Gemini Flash models and previewed its CodeMender code-security agent, aiming to reduce AI agent costs and automate vulnerability discovery. The centerpiece Gemini 3.6 Flash uses up to 17% fe…
Infinity, a startup founded by former Google Brain researcher Jeremy Nixon, raised $15 million in seed funding at a $100 million valuation to develop its AI agent Ignition, which automates the writing of low-level chip c…
A developer explains Kimi K3's KDA (Kimi Delta Attention) mechanism, a variant of linear attention that enables the model's 1-million-token context window. The post provides a 20-line Python example contrasting standard …
The US Treasury recovered and prevented more than $4 billion in fraudulent or improper disbursements in fiscal year 2024, a sixfold increase from $652.7 million in fiscal year 2023, driven by new pre-payment screening to…
Researchers have developed Sobek, a generated-CUDA backend for streaming equivariant tensor product convolutions that eliminates edge-sized intermediates, achieving speedups of 1.2× to 49.7× and reducing peak memory by u…