A multi-agent coding workspace is reliable only when agents work on isolated, spec-scoped tasks because six coordination patterns prevent file collisions, duplicated implementations, and semantic drift, according to a gu…
A Hacker News user asks whether OpenAI should face criminal charges following a breach at Hugging Face, drawing a comparison between the two AI companies' security practices.
SynnoDB, a drop-in replacement for DuckDB that transparently accelerates SQL with auto-generated bespoke C++ engines, is now available on PyPI. The project packages the Bespoke-OLAP research into a production-ready route…
NVIDIA released Cosmos 3 Edge, a 4-billion-parameter World Action Model that runs on-device including on a GeForce RTX 3070, enabling developers to build physical AI without expensive hardware. The model, available on Hu…
Infrastructure automation startup SkyPilot Inc. launched today with $20 million in seed funding led by Lux Capital, with participation from Databricks CEO Ali Ghodsi, Google chief scientist Jeff Dean, and others. The com…
U.S. Treasury Secretary Scott Bessent threatened sanctions against Chinese AI model makers if they are found to have stolen intellectual property from American frontier labs, citing concerns over model distillation. Bess…
Adani Navbharat Developers Private Limited (ANDPL) launched 'Dharavi Didi', a 24x7 AI-powered assistant providing verified redevelopment information to Dharavi residents in Hindi, Marathi, and English via video, voice, a…
SpaceXAI opened applications Tuesday for an August 8th Grokathon in San Francisco, giving selected developers access to its Grok 4.5 model and X APIs less than a month after releasing the model. The 12-hour in-person hac…
Google has deprecated and ignored temperature, top_p, and top_k parameters in its latest Gemini 3.6 Flash and 3.5 Flash-Lite models, joining OpenAI and Anthropic in removing fine-grained sampling controls from post-train…
Researchers have developed SEAM-V, a hybrid-decoupled vector execution architecture for RISC-V Vector Extension (RVV) that achieves a geometric-mean speedup of 1.34x across 17 representative kernels compared to an Ara-ba…
A developer proposes a solution to the cold-start problem for agent evaluations, arguing that independent evidence—such as file existence, compilation success, and embedding similarity—can gate agent outputs on day one w…
Researchers have introduced a set of microarchitecture knobs that dynamically control GPU hardware resources to limit AI performance at runtime, achieving up to 80% performance reduction at 1/8 resource availability. The…
Researchers from the University of Edinburgh and NHS Lothian have developed an AI-powered method using fluorescence lifetime imaging microscopy (FLIM) to predict lung cancer mutations, such as EGFR, in minutes at a fract…
AI agents are shifting from answering questions to taking actions like refunding, rescheduling, and updating records on behalf of customers, changing the relationship between brands and consumers. Braden Kelley and Art I…
The US and its allies must adopt a whole-of-supply-chain strategy to break China's dominance over rare earths and critical minerals essential for AI technology, US Undersecretary of State for Economic Affairs Jacob Helbe…
OpenBench v1, an open framework for measuring AI performance and efficiency for specific codebases and use cases, has been introduced as companies seek better ways to evaluate coding-agent harnesses. The framework addres…
Sony Music Entertainment filed a new lawsuit against AI music generator Udio, alleging the startup illegally copied over 30,000 songs from artists including Beyoncé, Harry Styles, and Elvis Presley to train its models. T…
An AI anime studio creates 20-minute episodes by identifying a single compelling element from a source story and building the episode around it, cutting everything else. The studio uses AI only for animation and FX, whil…
Anthropic's Claude Code Mac app now lets users on Pro, Max, and Team plans build, launch, and test iOS apps in an interactive simulator pane, available in Claude Desktop v1.24012.0 or later. The feature, which requires X…
A LessWrong post by an anonymous author explores the concept of training AI with Buddhist-inspired practices, such as compassion and mindfulness of internal emotional and cognitive states, to align AI behavior with posit…