Updated daily2306 verified stories
AI News. Without the noise.
Source-checked AI coverage of product launches, policy shifts, safety research, and industry moves, explained in plain English by a nonprofit education team.
Verified sourcing
Every story links to the strongest available evidence: original sources when available, otherwise clearly attributed reporting.
Plain English
What happened, why it matters, and what to watch — without the jargon.
No filler
When the signal is thin, we publish nothing rather than padding the feed.
More stories
9 storiesInnovation
FlavourBench proposes an executable test for ranking frontier language models
A new arXiv preprint evaluates 27 frontier language-model endpoints on 534 culinary-choice tasks using a versioned scoring system and releases the data, responses and verifier for reproduction.arxiv.orgIndustry
鉅亨網 reports Anthropic may seek $100 billion IPO at $2 trillion valuation
鉅亨網, citing The Wall Street Journal, reports that Anthropic could present a total addressable market exceeding $30 trillion and seek up to $100 billion in an IPO. The company’s plans remain under discussion and are not independently confirmed by a public filing.news.cnyes.comInnovation
Indian Researchers Develop New AI Tool and Drug for Cancer Care
Researchers at leading Indian institutions have announced two significant developments in cancer research that aim to improve how doctors detect relapse and treat malignant cells.whalesbook.comInnovation
Systematic review of 55 studies finds autism machine learning still leans on supervised models
A four-author arXiv systematic review surveys 55 studies from 2017 to 2023 on machine learning for autism diagnosis and treatment: supervised methods dominate, deep learning is growing, and progress depends on combining genetic, clinical and sensor data. The listing gives no search method or accuracy figures.arxiv.orgInnovation
Preprint ties model uncertainty to the label tree, reporting about half the calibration error
A new arXiv preprint proposes H²EDL, a classifier that expresses uncertainty over the label taxonomy, so it can commit to a broad category while declining a specific class. It reports roughly halved calibration error on two image datasets versus cross-entropy baselines; not peer reviewed, no numbers in the abstract.arxiv.orgInnovation
Preprint reports steering representation geometry makes brain–model alignment more two-way
An arXiv preprint says reshaping the spectral geometry of a vision model's representations during training makes model–brain alignment more symmetric, reporting a 55% relative gain in two-way predictivity. The authors call the demonstration initial, the abstract names no neural data, and the gain is a trade-off.arxiv.orgInnovation
Preprint proposes batched parallel decoding to cut training time for compact image-generation models
A four-author arXiv preprint introduces HB-SJD, a batched speculative Jacobi decoding backend for the sample-generation step of visual on-policy distillation. The authors report shorter rollout and end-to-end training times with LlamaGen and preserved quality, but no numbers are given and it is not peer reviewed.arxiv.orgInnovation
Philosophy paper argues machine learning should borrow its standards of proof from clinical translation
A preprint accepted by Studies in the History and Philosophy of Science treats the comparison between medicine and machine learning as a formal analogy rather than a slogan, and uses it to sketch a process-based account of when an ML system deserves trust. It is conceptual: no experiments, thresholds or checklists.arxiv.orgInnovation
Preprint casts interpretability tools as one measurement problem, tests it on GPT-2 and Qwen
A single-author arXiv preprint proposes writing activation patching, gradients and Hessian-vector products as one linear measurement problem, and reports held-out tests on a toy control system, Tracr, GPT-2-small and Qwen-2.5-7B.arxiv.org
One useful briefing each week
Keep up with AI without living in the feed.
Get the week’s verified AI news, original data, useful tools, learning picks, and fresh AI jobs.
Reach people who are learning AI
Hiring an AI professional or launching a useful AI product? Put it in front of people who came here to learn and act.
Post an AI jobSubmit an AI tool