Trapper Clam
397 posts

Trapper Clam
@oldclam
I am as old as the Atlantic; I've been here for millions of years. I was here when asteroids pummeled the Earth and killed most life. I am a fossil.





@BlackthorneAI @Pizza_Later Yep. If you think otherwise, you are ignorant of what actual brains do.









A new paper from Stanford adds further, incontrovertible evidence that LLMs memorize training data. The authors showed that copyrighted works can be extracted from every LLM they tried: GPT-4.1, Gemini 2.5 Pro, Grok 3, and Claude 3.7 Sonnet. For example, Claude 3.7 Sonnet reproduced *95.8%* of Harry Potter and the Philosopher's Stone. "We find that is possible to extract large portions of memorized copyrighted material from all four production LLMs". Some in AI, and even some courts, have claimed that AI models don't memorize training data. This is simply false. They do, as has been repeatedly shown. arxiv.org/pdf/2601.02671…







Hit Christmas gift this year: Stickerbox. Press button, describe something, AI generates an image and prints on a sticker. Great product design. We’ve printed hundreds of stickers so far. Made by one of my YC Summer 2009 batchmates!


AI didn't steal anything. It just did what we all did. It learned from others, looked at pictures, watched videos, read articles, studied code, and now it's creating new things with what it has learned. And now, it's sharing with you, for free. Enjoy.

























