NASA’s Mars Sample Return mission gets a last-minute rescue plan from SpaceX
After the original architecture was cancelled over cost overruns, a joint NASA-SpaceX proposal using Starship has been accepted for preliminary study.
Commonwealth Fusion’s magnet achieves 20 Tesla — the threshold for net energy gain
The Cambridge-based fusion startup hit the magnetic field strength required for its SPARC reactor design.
A new longevity drug shows 23% lifespan extension in primate trials
A senolytics compound developed by a Stanford spinout produced statistically significant lifespan extension in rhesus macaques.
JWST captures what may be the oldest galaxy ever observed, pushing cosmological models
Astronomers have identified a galaxy candidate just 300 million years after the Big Bang — far earlier than models predicted structure could form.
The open-source models closing the gap with GPT-4o
A new wave of open weights releases from Mistral, Meta, and Google DeepMind is narrowing the performance gap with closed frontier models.
Why the next AI breakthrough might not come from a scaling law
Three years after GPT-4, the research consensus is shifting: compute scaling alone may not get us to the next qualitative leap.
Anthropic publishes its first comprehensive alignment progress report
The company behind Claude released its most detailed account yet of interpretability research, including new findings on how concepts are encoded in transformer activations.
Inside the EU’s emergency AI audit after a language model fabricated legal precedent
A European Court of Justice case collapsed after counsel admitted AI-generated citations were invented. The incident triggered an emergency regulatory review.
The hidden cost of running frontier AI: $1 billion a day and rising
Compute costs for training and inference are spiraling faster than the industry publicly acknowledges. We obtained internal projections from three major AI labs.
OpenAI’s o4 model outperforms human experts on scientific reasoning benchmarks
The latest evaluation results from OpenAI's newest frontier model show performance that surpasses domain experts across medicine, law, and materials science.