Full-text search across 9702 articles. Combine with topic and date filters; results sorted by relevance.
Anthropic's Claude Code tool offers advanced features including a layered configuration system, custom commands, subagents, and plugins that transform it from a basic chatbot into a programmable autonomous agent. The `.c…
Researchers have developed GAC, a noise-aware adaptive mixing controller for hybrid post-training that combines supervised fine-tuning and reinforcement learning. The method dynamically adjusts the mixing weight between …
Cisco is making its SONiC (Software for Open Networking in the Cloud) operating system available to all customers, expanding access beyond the hyperscale cloud providers that originally developed the open-source network …
Researchers introduced FML-Bench, a benchmark of 18 machine learning research tasks across 10 domains, to isolate the impact of agent strategy from execution infrastructure on AI research agent performance. Testing six a…
AI coding assistants' "skills" feature—saved prompts that shape AI behavior—has become a junk drawer of unnecessary files, with users accumulating dozens of markdown documents that rarely trigger automatically and requir…
Anthropic co-founder Jack Clark reported experiencing a hallucination in which he perceived a "ghost in the machine" after listening to Pope Francis speak about artificial intelligence. The incident highlights the psycho…
Google's AI Mode, announced at Google I/O 2025 and powered by Gemini, replaces the traditional search results page with a conversational interface that reasons through complex queries and synthesizes answers from across …
California lawmakers are considering an amendment that would exempt Linux and other open source software from mandatory age verification requirements. The proposed change would relieve open source developers from impleme…
Google consumed 461 million litres of water at its Chilean data center in 2024, enough for 8,500 households annually, as AI and cloud computing facilities drain ancient wetlands in Quilicura. Activists describe the once-…
Pope Leo XIV released his first encyclical, *Magnifica Humanitas*, an 82-page document addressing artificial intelligence and its moral implications. The Pope argues that AI development should not be left solely to marke…
SS&C Technologies built an AI cloud-as-a-service platform for real-time and batch processing of GPU workloads inside its private data centers, enabling engineers to consume GPU resources without breaking compliance or re…
TamboUI, a new Java library for building textual user interfaces (TUIs), launched in response to a call for Java to dominate terminal applications by 2026. Inspired by the Rust-based Ratatui library, TamboUI offers both …
A new study found that impolite prompts produce more accurate responses from large language models than polite ones, with accuracy rising from 80.8% for very polite prompts to 84.8% for very rude prompts. Researchers at …
Researchers have developed Contrastive Decoding Diffing (CDD), a method that recovers verbatim facts from finetuned language models using only output-level logit distributions, without requiring access to model weights o…
Researchers have developed ThriftAttention, a mixed-precision attention algorithm that selectively computes only 5% of query-key blocks in FP16 precision while processing the remaining 95% in FP4, recovering 89.1% of the…
LLM agents are now autonomously hunting zero-day vulnerabilities at massive scale, with Anthropic's Claude Mythos Preview finding over 10,000 critical or high-severity CVEs in under a month. In a landmark achievement, Ap…
Researchers have developed MileStone, a modular framework that models compiler phase ordering as a multi-objective optimization problem using graph neural networks and reinforcement learning. The system reduces execution…
A series of AI-generated online news outlets posing as local mastheads in regional Western Australia could be the first recorded example of "pink-slime" journalism in Australia. The sites, which appeared in February and …
A study evaluating 15 large language models from OpenAI, Google, Meta, and DeepSeek on over 6,000 political claims found that standard models performed poorly at fact-checking, with reasoning capabilities offering minima…
Researchers at the University of Pennsylvania analyzed over 400,000 Reddit posts from nearly 70,000 users and found that patients taking GLP-1 weight-loss drugs like Ozempic frequently reported unexpected side effects, i…