Don’t trust AI companies with your content. Verify them instead
(feeds.feedburner.com)
31.
32.
What happened when Meta tried to collect employee data
(feeds.feedburner.com)
35.
Moonshot serves Claude instead of Kimi and collects exchanges for model training
(news.ycombinator.com)
36.
37.
38.
Google will buy half the electricity of a nuclear power plant
(news.ycombinator.com)
39.
Tell HN: OpenAI keeps re-enabling the 'allow training' setting
(news.ycombinator.com)
40.
Another researcher says OpenAI trained on conversations, then claimed breakthrou
(news.ycombinator.com)
41.
Mathematicians want proof OpenAI didn't use their work
(news.ycombinator.com)
42.
Mathematicians want proof OpenAI didn’t use their work
(theverge.com)
43.
Training a 3.8B LLM to 0.384 CORE for $998
(news.ycombinator.com)
44.
Training a 3.8B LLM to 0.384 CORE for $998 – Hugo Vergnes
(news.ycombinator.com)
45.
46.
47.
48.
49.
End-to-end infrastructure for training and inferencing open weight models
(news.ycombinator.com)
50.
“Next-token predictor” is the wrong mental model for LLMs
(news.ycombinator.com)
51.
"Next-token predictor" is the wrong mental model for LLMs
(news.ycombinator.com)
52.
Stop Thinking of LLMs as Next-Token Predictors
(news.ycombinator.com)
54.
55.
57.
58.
How Railroad Crossings Work
(news.ycombinator.com)
59.
60.
Can I opt out of my input or output data being used for training?
(news.ycombinator.com)
Today's top topics:
openai
anthropic
apple
ai safety
google
iphone 18 pro
artificial intelligence
ios 27
dario amodei
nvidia