Three sites made 215,000 pages targeting AI, and Perplexity cited them
A study of 380 software categories found that six in ten citations pointed to obscure sites — and three of those sites were built purely to target AI
310
Hyunkook Kim
김현국
Meta's Muse Spark 1.3 beats GPT-5.6 and Opus 5 on coding benchmarks
It leads on long-context and coding scores, but still trails Opus 5 across general knowledge work
130
Hume AI ranks human-like voice third in its evaluation framework
CEO Andrew Ettinger revealed the internal CLEAR scoring criteria in a video
50
Google unveils Gemini 3.8 Flash, tuned for coding and agentic work
The new model builds on Gemini 3.7 Flash, ships across six deployment paths, and supports up to 1 million tokens of context
220
Databricks Publishes 'Big Book' Guide To Running AI Agents In Production
A six-chapter guide for moving AI agents from pilot to production, with DXC and ICE case study numbers included
130
Ai2 outlines five limits of AI-assisted scientific research
At an event marking an expanded partnership with Providence Swedish, Ai2 addressed where AI still falls short on judgment and verification in science
70
Anthropic adds graduate and PhD tracks to Campus Ambassadors program
Anthropic opens three tracks — undergraduate, graduate, and PhD — each offering a $3,600 grant, with applications closing September 12
8600
World Labs' Atlas Generates and Reconstructs 3D Worlds From a Few Photos
Fei-Fei Li's World Labs has rolled generation, reconstruction, and robot simulation into a single model called Atlas
220
Ai2 finds BBQ safety benchmark actually measures reasoning ability
An audit of 16 benchmarks and roughly 34,000 items found many "safety" scores were actually tangled up with reasoning skill
110
12.5 Gigawatts on Sheep Pasture and Coal Country — China's Real AI Hub Isn't Beijing
Two hours by train from Beijing, roughly 100 data centers have sprung up in Ulanqab, Inner Mongolia. Cheap electricity, cold winters, and surplus wind power have drawn China's AI companies to this plateau — but underneath sit two unresolved problems: water scarcity and coal. One city tells you how China's AI infrastructure math actually works.
120
Open-Weight AI Becomes Cheapest Shield and Easiest Spear in Same Month
A single round of fine-tuning was enough to strip the safety guardrails off 21 open-weight models.
320
Claude Autonomously Designs Proteins, Nearly Half Bind in Lab Tests
Anthropic's Claude designed proteins on its own in 24 hours, given only a target — and when two independent labs tested them, nearly half actually bound
1600