631.
633.
635.
Claude Fable 5 promotion extended after backlash over early cutoff
(androidauthority.com)
636.
Judicious use of LLMs could speed up progress in the social sciences
(feeds.nature.com)
637.
Large language models can predict the results of social science experiments
(feeds.nature.com)
638.
Silent speech with ultrasound
(news.ycombinator.com)
639.
641.
642.
643.
Why the rise of open source AI isn’t hurting Anthropic … yet
(techcrunch.com)
644.
645.
646.
Anthropic Caught Secretly Spying on Users
(futurism.com)
647.
648.
Reducing Doom Loops with Final Token Preference Optimization
(news.ycombinator.com)
649.
650.
651.
Show HN: Frugon – Find which LLM calls a cheaper model could handle (local, MIT)
(news.ycombinator.com)
652.
653.
654.
655.
Small AI Models Gain Traction In places with unreliable networks
(news.ycombinator.com)
656.
The ‘first’ AI-run ransomware attack still needed a human
(techcrunch.com)
657.
658.
Ternlight – 7 MB embedding model that runs in browser (WASM)
(news.ycombinator.com)
659.
Small AI Models Gain Traction Around the World
(slashdot.org)