News on AI, science, space, and tech.
Sourced, checked, and free of hype.
Topics
- AI4 postsModels, tools, and research shaping artificial intelligence.
- Science1 postDiscoveries and ideas from labs and fieldwork.
- Space1 postMissions, telescopes, and the exploration of the cosmos.
- Tech1 postHardware, software, and the industry behind them.
- Insight4 postsAnalysis and context on what the news means.
Featured
AI Is Proving Math Faster Than People Can Check It. Who Decides What Counts as Solved?
OpenAI's 2026 math claims range from a checked result to an unreviewed pile. Here is what they mean for mathematicians and for how we trust new knowledge.
OpenAI Posts 722 AI-Written Math Papers. Three Were Pulled the Next Day.
OpenAI released hundreds of math results made by an unreleased model. A sign error led to three withdrawals, and many proofs are still unchecked.
Genie 4 Builds Worlds You Can Walk Through. Is That a Path to AGI?
Google's Genie 4 makes real-time worlds with sound. Does simulating a world mean an AI understands it? Here is what Google, critics, and researchers say.
More articles
11 articles · Most recentAI Can Now Study and Rebuild Software. What Do REA and Craft Anything Change?
REA lets AI agents take software apart. Craft Anything goes further and rebuilds it. Here is what they can do, who is affected, and what experts say.
The AI Race Shifts to Cost and Action. Is AGI Any Closer?
Today's launches stress low cost and real work. Lab chiefs say AGI is a few years away, while many researchers doubt it. Here is what each side would need to prove.
A New X-ray Test Adds Evidence for a Third Kind of Magnetism
Researchers used polarized X-rays on manganese telluride and saw a signature expected for altermagnets, a magnet class first reported in 2024.
NASA Details Artemis III: A Two-Week Docking Test in Earth Orbit
Artemis III will not land on the Moon. In 2027 it will test docking with two commercial landers in Earth orbit, before a landing attempt on Artemis IV in 2028.
Anthropic Releases Claude Fable 5.2 and Mythos 5.2 at Lower Prices
Anthropic's new top models run up to 30% faster and cost 20% less. Fable 5.2 is open to all users, while Mythos 5.2 is limited to vetted teams.
AI Is Pushing Memory Prices Up, and Phones Are Feeling It
DRAM revenue nearly quintupled in a year while shipments barely grew. Prices drove the jump, and chipmakers such as Qualcomm are passing costs on to phone makers.
Meta Launches Muse Ember, a Low-Cost Model for Repetitive Agent Work
Meta's smallest Muse model costs $0.10 per million input tokens and targets the repetitive steps inside AI agents. Its benchmark results are company-reported.
NVIDIA Unveils GR00T 2, a Robot Model That Learns From Human Video
NVIDIA's GR00T 2 learns dexterous skills from a few demonstrations. Open weights are promised for December, and the benchmark numbers come from NVIDIA.
