AI Research news and analysis on Web Pulse: 8392 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Anthropic has discovered a method to peer into its Claude AI model's internal reasoning, revealing a hidden layer of neural patterns called J-space that can detect when the model knows it is being tested or is hiding dec…
Approximately 80% of US startups using open-source AI have adopted cheaper Chinese models from DeepSeek, Alibaba, and Moonshot AI, according to Andreessen Horowitz. The shift, driven by costs as low as one-eighth of US a…
LG AI Research presented its Exaone model's industrial applications at the International Conference on Machine Learning in Seoul, including AI-discovered materials like a hair loss treatment ingredient and a cooling flui…
Illinois Governor JB Pritzker signed SB 315 on July 6, making Illinois the first US state to require mandatory annual independent audits of frontier AI developers. The law targets companies like OpenAI, Anthropic, Google…
OpenAI announced the public launch of its GPT-5.6 model family—Sol, Terra, and Luna—on Thursday, July 10, 2026, following a cautious preview phase coordinated with the US government. The naming overlap with collapsed cry…
A study of about 1,080 job postings at OpenAI and Anthropic reveals the two AI labs are diverging in their approaches to agentic AI, which poses a security risk as AI agents' behavior can drift over time, evading standar…
Researchers propose Patch Knowledge Transfer (PKT), a knowledge distillation framework that enables efficient AI-generated image quality assessment by transferring knowledge from a complex teacher model to a lightweight …
Researchers proposed a controllability-observability framework for compressing deep neural networks by viewing them as nonlinear dynamical systems. The method achieved up to 83% parameter compression on CIFAR-10 with neg…
Researchers introduced PatchOptic, an optic-inspired interface for shared-state LLM workflows that enforces contracts between local updates and global validity through projected reads and verified structured patches. Eva…
Researchers introduced Self-Review Reinforcement Learning (SRRL), a training framework that embeds a self-review step into each RL episode to help large language models learn from sparse feedback. SRRL outperformed stand…
Researchers introduced SPANUQ, a lightweight probe for span-level uncertainty quantification in large language models, achieving 10-20x faster inference than sampling-based methods while outperforming them in uncertainty…
Researchers replaced full-corpus injection in a legal-document analysis system with two structured retrieval modes, achieving comparable accuracy while reducing token usage by up to 29.9x and costs by 25%. The study foun…
Researchers introduced Prompt-to-Paper, a multi-agent AI system that generates verifiable bioinformatics manuscripts by grounding claims in 60-100 cited papers, executing real computational experiments, and scoring quali…
Researchers introduced CSTutorBench, a benchmark to evaluate small language models as tutors for block-based programming in VEX VR robotics. Testing 11 models revealed that instruction-tuning approach predicts tutoring q…
Researchers demonstrate that moving memory retrieval inside the agent loop reduces latency from hundreds of milliseconds to ~100 microseconds, eliminating redundant actions and improving recall. In tests across GPT-5-cla…
Researchers tested whether large language models can generate synthetic consumer data for projective techniques, finding substantial overlap with human responses in broad topics but differences in style and diversity. Th…
A synthesis of 27 papers (2023-2026) spanning 19 benchmarks reveals six recurring failure clusters in large language model agents, including tool-use errors, planning failures, and safety issues, with failures compoundin…
Researchers introduced EquiFiLM, a lightweight extension that adds continuous external conditioning to equivariant foundation machine learning force fields via per-layer Feature-wise Linear Modulation. Applied to charged…
Researchers introduced NapMem, a framework that transforms long-term user memory into a structured action space for conversational agents, enabling active navigation across memory granularities. Trained with reinforcemen…
Researchers introduced Onnes, a physics-grounded multi-agent LLM simulator for cryogenic fault diagnosis in quantum computing infrastructure. In controlled evaluations, a zero-shot LLM panel matched a supervised ML class…