The Archive

ZML, a hot French AI startup endorsed by Turing Award winner Yann LeCun, has now released ZML/LLMD, software that could make running AI less costly.

AI chip maker SambaNova has raised at an $11B valuation months after Intel was rumored to be trying to buy it for about $1.6 billion.

"HalluSquatting" weaponizes LLMs' inability to say "I don't know."

Lilian Weng summarizes 35 papers on harness engineering for RSI; meta-analysis of recent research without new findings.

OpenAI releases GPT-Live, a new voice model generation for natural human-AI interaction in ChatGPT Voice.

The new image-generating model has numerous use cases, including advertising, decorating and creator-based opportunities.

Squeeze on Rust Belt electricity bills threatens Trump’s manufacturing plan.

Meta is launching the first AI image generation model made by its Superintelligence Labs division. The Muse Image model now powers the image-making tools across the Meta AI app, Instagram, and WhatsApp, and it's coming soon to Facebook and Messenger, according to an announcement on Tuesday. It's part of the growing Muse family of AI models that replace Meta's Llama lineup. Alexandr Wang, who Meta hired to head up its Superintelligence Labs last year, says on Threads that Muse Image is "agentic," meaning it works with its Muse Spark large language model "to reason through your prompt, search t...

Open-source models’ success isn’t coming at the expense of frontier labs. Instead, they each seem to capture two phases of the same life-cycle.

Microsoft is the latest Silicon Valley giant to cut back on its AI spending.

sqlite-utils 4.0 adds schema migrations, nested transactions, and compound foreign key support to SQLite tooling.

Discord has acknowledged that a bug in its AI moderation system mistakenly banned more than 8,000 users over the past two months, after harmless images—including spreadsheets, chessboards, game textures, as well as white and gray transparent backgrounds—were incorrectly flagged as harmful content. The company confirmed that the issue had been affecting accounts since May, with […]

ELSA3D introduces elastic semantic anchoring for unified 3D understanding and generation, structuring language-geometry reasoning across matched abstraction scales.

Graph Convolutional Attention analyzes spectral properties of attention-based graph denoising, showing linear attention limitations for graph diffusion tasks.

Rethinking Indic AI examines cultural-linguistic preservation and inclusion challenges for underrepresented Indian languages in AI systems.

Unsupervised dependency parsing via network science methods applied to non-human sequences without gold standard evaluation.

Starting Tuesday, Anthropic's Claude Cowork AI platform will be available on mobile and web for the first time. The expanded access is rolling out first to Max subscribers and coming to Claude users on other plans "in the coming weeks." Claude Cowork was previously only accessible through the Claude desktop app for macOS and Windows, but now users on iOS and Android can also use it. However, Anthropic says the "full experience" for Cowork will still be on the desktop app, including features like local file access. Cowork sessions will also now run in the cloud by default, so you can continue ...

Hierarchical Acoustic-Semantic Modeling addresses modality interference in full-duplex Spoken Language Models via separation and semantic coherence techniques.

GraphBU proposes graph-native MILP instance generation using block units that preserve structural coupling for solver development.

Large Cancer Assistant (LCA) is a model-agnostic orchestration framework for multimodal clinical decision support in oncology.

Life Style Levels uses satellite imagery and morphological indicators to delineate socioeconomic variations in Indian urban areas.

RSF-GLLM bridges semantic gaps in multi-hop knowledge graph QA via recurrent soft-flow reasoning and decoupled LLM generation.

DepthWeave-KV compresses long-context KV cache via token-adaptive cross-layer factorization, improving inference efficiency.

VAORA method improves VLM reasoning alignment in physical interactive tasks by anchoring chain-of-thought to visual context via dual reward signals.

FreqDepthKV compresses KV cache in long-context LLM inference by factorizing layers into shared low-frequency and sparse high-frequency components.

FootsiesGym is an open-source fighting game benchmark for two-player imperfect-information RL research with vectorized simulation.

DynaKRAG learns unified control policies for multi-hop RAG operations including retrieval, reformulation, critique, and sufficiency judgment.

NAICS-GH dataset maps 6,588 GitHub repos to North American Industry Classification System sectors for innovation geography studies.