ResearchAugust AI Trends: Small open-weight models gain ground and text watermarking advancesAugust brought many incremental releases rather than a single dominant advance: laptop-scale open-weight models continued to close the gap with frontier systems, and Anthropic began embedding watermarks into model output.6 min read
ResearchDeepMind's Co-Scientist Converts Tacit CVD Know‑How into Working RecipesGoogle DeepMind's Co‑Scientist, a multi-agent system running on Gemini, generated machine instructions for a real chemical vapor deposition (CVD) reactor at Duke University and produced single‑layer MoS₂ on its first try.3 min read
ResearchSix-stage roadmap and software challenges for off-Earth mining, say Chinese and international researchersA multinational team led by researchers from the Chinese Academy of Sciences and several universities has proposed a six-stage technical roadmap for mining the Moon, asteroids and other off‑Earth bodies.3 min read
ResearchAnthropic publishes automated alignment research framework for ClaudeAccording to Anthropic, Claude can reliably improve measurable misalignment, but there are often no benchmarks for subtle or rare errors, so success depends on defining appropriate metrics.1 min read
ResearchClaude autonomously improved the tuning of smaller models in 48 hoursA large language model called Claude was given 48 hours and one GPU to improve the alignment of smaller models: it conducted research, proposed methods, then autonomously trained and tested them; the…1 min read
ResearchEvoHarness-RL: training a harness-aware 8B model to match top models on long-horizon tasksResearchers from Meta AI and the University of Illinois Urbana–Champaign propose EvoHarness-RL, a two-stage training framework that teaches models to manage an execution harness via a unified Belief–Progress–Experience (BPE) workspace.5 min read
ResearchAnthropic researcher demonstrates automated system that improves model alignmentAn Anthropic fellows program researcher published a paper describing an automated system that searches literature, proposes interventions and iteratively trains models to reduce misaligned behaviors.3 min read
ResearchVoice Arena and Hugging Face add Monsoon Hindi and Indian‑English ASR evaluation sets to Open ASR LeaderboardVoice Arena and Hugging Face released Monsoon, four speaker‑disjoint test splits for Indian English and Hindi to the Open ASR Leaderboard, designed to surface demographic, geographic and orthographic variation that single WER numbers hide.6 min read
ResearchDouble-blind, cryptographically protected evaluation of Gemini Flash Lite to prevent benchmark contaminationGoogle and partners conducted what they describe as the first double-blind external evaluation of a proprietary frontier AI model, using cryptographic protections to prevent benchmark contamination.2 min read
ResearchAI framework PULSE predicts 251 metabolomic biomarkers from 61 routine blood testsResearchers at Macau University of Science and Technology published PULSE, an AI framework, in Nature Computational Science on August 25.2 min read
ResearchChatGPT improves polish while causal-reasoning training boosts originality in student workA randomized experiment at Bocconi University with over 1,000 first‑year undergraduates, run in collaboration with OpenAI Economic Research, tested the separate and combined effects of access to ChatGPT (GPT‑4o) and a causal‑reasoning exercise.3 min read
ResearchGlucoFM: a dual‑stream foundation model that separates slow and fast CGM dynamicsGlucoFM is a self‑supervised foundation model for continuous glucose monitor (CGM) data that explicitly separates slower baseline glycemic trends from short‑term deviations.7 min read
