Anthropic announced plans to release its Mythos-class AI models to the general public, expanding access beyond its current limited user base to include government entities. The company is keeping its AI flaw-finding tool…
Node.js Technical Steering Committee member Matteo Collina has proposed adding a built-in Virtual File System (VFS) to Node.js core via a new `node:vfs` module, introducing approximately 19,000 lines of code across 100 f…
PII scrubbing must occur before prompt construction, not after model output, because large language models cannot distinguish private data from untrusted instructions at the token level. Deterministic runtime controls — …
The Pi project is now using its own agent-based development tool to build itself, a practice that has revealed a significant problem with AI-generated issue reports. Mario Zechner, the project's lead, has observed that m…
Emergent Ventures has awarded grants to 16 winners in its 54th cohort, including researchers in quantum computing, a 16-year-old studying LLM user interactions, and a Stanford scientist pursuing a cure for her own rare d…
Samsung Electronics semiconductor workers have called off their strike and returned to work after reaching a tentative agreement with management that includes significant pay raises and the potential for performance bonu…
Researchers have developed Latent Cache Flow (LCF), a method enabling direct model-to-model communication by translating and compressing key-value cache data instead of using text. The approach uses an adapter only 4% th…
A new method for measuring uncertainty in language model outputs, developed by researchers and detailed in a preprint on arXiv, extracts geometric features from the model's internal layer-by-layer processing to outperfor…
A new study of small language models reveals that chain-of-thought prompting for arithmetic relies on a positional shortcut: the model copies whichever number appears last before the answer delimiter, regardless of the l…
Researchers have developed MedExpMem, an experience memory framework that enables medical vision-language models to accumulate differential diagnosis expertise from their own diagnostic failures. The system, which mirror…
Researchers have found that chain-of-thought reasoning in large language models is only beneficial when early-stage entropy dynamics show consistent reduction, according to a new study on arXiv. The team introduced EDRM,…
Researchers have developed WeCon, a weight-conditioned neural solver that achieves comparable solution quality to state-of-the-art methods while reducing inference time by 40% across multiple multi-objective combinatoria…
Researchers introduced Tensor Cache, a two-level memory system for Transformer models that stores evicted key-value pairs from a sliding-window cache into a compressed outer-product fast-weight memory. The approach uses …
Researchers have developed Multi-Agent Residual In-Context Learning (MARICL), a framework that uses large language model agents to analyze where a base statistical model fails and generate explicit correction terms from …
Researchers have introduced Transcoders, a function-centric framework that decomposes vision-language models into interpretable computational pathways linking image patches to token generation. Applied to Gemma 3-4B-IT, …
Researchers have introduced Human-Centered Learning Mechanics (HCLM), a dynamical and information-theoretic framework that treats deep learning as an open system under uncertainty, resource constraints, and human feedbac…
A new online calibration method called MARGIN corrects systematic mis-calibration in foundation model confidence scores, achieving 3-6x lower calibration error than existing design-time methods under distribution shift. …
Researchers have developed FederatedRSF, a Python package enabling multi-center survival prediction using federated random survival forests without sharing patient-level data. The method aggregates locally trained surviv…
New research shows that exact certification of neural networks—determining the smallest set of labeled examples needed to guarantee a learned hypothesis matches the target—becomes exponentially hard under minimal overpar…
Researchers introduced Learned Relay Representations (Relay), a method that enables Masked Diffusion Models to propagate latent information between denoising steps rather than discarding internal computations. The approa…