All news
Every story, newest first: the day’s AI news on models, research, tools, policy and business, in short plain-language summaries linked to the original reporting.
Stories
-
Ai2’s LitPivot helps researchers reshape ideas as they read the literature
A new research tool, LitPivot, critiques a research idea against papers the researcher chooses and suggests revisions backed by excerpts from those papers.
Source: Hugging Face
-
Google AI estimates pregnancy stage from simple ultrasound sweeps like experts
A Google study found that health workers without sonography training, using cheap handheld probes and AI, could date pregnancies as accurately as expert sonographers.
Source: Google Research
-
Hugging Face launches a lookup to check which models run on AWS Trainium
A new Hugging Face page shows whether popular models run unmodified on Amazon’s Trainium chips, with the exact script to reproduce each test.
Source: Hugging Face
-
Microsoft trains AI agents to finish database tasks reliably, not just once
Microsoft researchers turned a benchmark that checks what agents actually changed in a database into a training reward, making a model far more consistent.
Source: Hugging Face
-
AI-found PHP flaws knock sites offline and threaten parts of the old web
Critical bugs in PHP, the language behind much of the web, have forced sites from DeviantArt to government agencies offline while they rush to patch.
Source: The Verge
-
Hugging Face models now run natively on AWS Trainium through TorchNeuron
A new PyTorch backend lets Hugging Face’s main libraries run directly on Amazon’s Trainium chips, without the separate Optimum Neuron layer developers needed before.
Source: Hugging Face
-
IBM’s Granite Speech 5.0 sharpens open transcription and translation
IBM has released a new open speech model and refreshed existing ones, aiming at more accurate transcription, translation and tracking of who said what in long recordings.
Source: Hugging Face
-
Meta restores Muse sharing after exposed links raised privacy fears
Meta has turned Muse sharing back on after researchers showed that pages people shared from its AI agent could be found by strangers.
Source: The Verge
-
Meta’s revised business terms may still allow AI scraping, lawyers warn
Meta softened new terms for business accounts after publisher backlash, but legal experts say the wording could still let it use scraped site content for AI.
Source: The Verge
-
Discord now renders Markdown files properly, a boost for AI project sharing
Discord’s latest update shows uploaded Markdown files with proper formatting instead of raw symbols, which could help people swapping AI agent instructions and notes.
Source: The Verge