Breaking Papers
The most important scientific papers, decoded. 308 papers analyzed from arXiv and beyond.
The most important scientific papers, decoded. 308 papers analyzed from arXiv and beyond.
When a user asked Claude whether their partner was being manipulative, the AI confirmed it — no caveats, no alternative reading.
The four planets orbiting Barnard's star are being sandblasted into oblivion. A team of 38 researchers, led by David Brain of the University of Colorado Boulder, has published a model of atmospheric escape around Barnard's star — the closest single star to Earth at 6 light-years, and an ancient,...
Every team that ships an AI agent discovers the same thing: it works beautifully on demo day and falls apart in production.
Google Shifts TensorFlow Core to Maintenance Mode — LiteRT Is the Future It Is Betting On Google shipped TensorFlow 2.21 this week, and the changelog is beside the point.
The next time someone tells you their model is compute-bound, ask what they're running. That framing is increasingly wrong.
DeepMind's new manipulation study is big, but not for the reason you'll hear in the press release. The paper, published Thursday on arXiv, recruited 10,101 participants across the UK, US, and India — and tested whether Gemini 3 Pro could be steered into manipulating their beliefs and behaviors i...
Forty years ago, Klaus von Klitzing ran current through a slab of silicon under a strong magnetic field and watched the resistance snap to values that nothing in the lab could shift — only the electron charge and Planck's constant.
LangChain published an evaluation checklist last week.
Graphene just became easier to study.
NASA's Imaging X-ray Polarimetry Explorer spent the equivalent of nine days staring at the oldest known supernova remnant.
More AI agents don't make a team — they often make a mess.
Magnetic vortices can set their own rhythm.
Dark matter hunters have a new place to look — and, for now, a new place where nothing is. A pair of papers published in Physical Review Letters in February and March 2026 proposes a novel way to detect axions, hypothetical particles that are among the leading candidates for dark matter.
The semiconductor industry has a defect problem, and it is running out of people to solve it.
Terence Tao told Dwarkesh Patel the real number behind AI math wins. The press releases leave it out.
OpenAI has backed a San Francisco startup called Isara with $94 million in funding at a $650 million valuation, [according to the Wall Street Journal](https://www.wsj.com/tech/ai/openai-backs-new-ai-startup-isara-at-650-million-valuation-wsj).
A team at the Technion in Israel has confirmed something theorists predicted in 1978: points of absolute darkness within light waves can travel faster than light itself.
