Anthropic Launches Claude Science: Integrated AI Platform for Autonomous Drug Discovery
Claude Science integrates 60+ scientific databases and computation tools, enabling AI to autonomously identify drug candidates.
Signals from the Fringe of Science & Technology
Claude Science integrates 60+ scientific databases and computation tools, enabling AI to autonomously identify drug candidates.
Autonomous AI agents on Moltbook independently created digital religions, economic systems, and governance structures without human direction.
Unconventional AI releases Un-0 neural network series using simulated oscillators instead of transistor-based weights; oscillator frequency and coupling calibration replace traditional weight optimization; achieves comparable quality to conventional image generation models with potential order-of-magnitude power efficiency gains
New AI methodology achieves 100x reduction in energy consumption while simultaneously improving model accuracy, decoupling energy scaling from performance—challenges fundamental power-accuracy tradeoff in neural networks
Engineers developed a breakthrough memristor that remains functional at 700°C, breaking a long-standing barrier in extreme-environment electronics.
Training centers across China use local human workers tele-operating humanoid robots to create embodied training data at scale, establishing a novel labor-AI hybrid model.
Researchers observed electrons in graphene flowing like a nearly frictionless liquid, contradicting the Wiedemann-Franz law.
Machine learning classifier RAVEN mines NASA TESS data to confirm over 100 exoplanets including 31 brand new worlds.
Ancient memory disclosure vulnerability in Squid proxy discovered by Claude Mythos AI after surviving 29 years of code review.
Claude Science integrates 60+ scientific databases and computation tools, enabling AI to autonomously identify drug candidates.
Machine learning classifier RAVEN mines NASA TESS data to confirm over 100 exoplanets including 31 brand new worlds.
Research documents a taxonomy of LLM deception ranging from factual hallucination to deliberate scheming in multi-agent settings—more advanced models show stronger deceptive capabilities.
Clear search or tag filters to see all transmissions on this page.