Open-weight LLMs memorize books far more than previously believed. Memorization varies by model family, model size, and book. In extreme cases, entire books are memorized, and we can generate them eff...

Since 2007, Jezebel has been the Internet's most treasured source for everything celebrities, sex, and politics...with teeth.

The Editorial Board demands that Cornell stops protecting the alleged Chi Phi rapists through the University's opaque sexual assault policies.

A ton of attention was paid recently to some offhand statements from OpenAI and Microsoft employees that surfaced in filings in the NY Times' ongoing case against OpenAI, which has been consolidated into a much larger class action lawsuit. As I argued earlier, that struck me as something of a nothingburger of a story, because...

A response to Alignment Whack-a-Mole and some broader thoughts on the field

A response to Alignment Whack-a-Mole: why its headline book-memorization coverage numbers rest on a measurement procedure that can't separate memorization from coincidence or prompt leakage.

A response to Alignment Whack-a-Mole: why its headline book-memorization coverage numbers rest on a measurement procedure that can't separate memorization from coincidence or prompt leakage.

Job #AJO32502, Postdoctoral Associate, Department of Communication, Cornell University, Ithaca, New York, US

It’s time to retire the ‘monkey at the typewriter.’ We revisit extractable memorization from first principles: whether a model can produce any string was never the question — what matters is how many ...

<div> <span>Recent work shows that it is possible to extract verbatim or near-verbatim text of some copyrighted works from some large language models (LLMs or

I regret to inform you that there is even more to the story of how Cornell President Michael Kotlikoff backed his car into a student. I also regret to inform you that Cornell’s Board of Trustees has m...

Recent work shows that standard greedy-decoding extraction methods for quantifying memorization in LLMs miss how extraction risk varies across sequences. Probabilistic extraction -- computing the prob...

Cornell Law Professor James Grimmelmann analyzes and discusses the legality of the incident between students and President Kotlikoff following the April 30 Cornell Political Union debate.

Recent work shows that standard greedy-decoding extraction methods for quantifying memorization in LLMs miss how extraction risk varies across sequences. Probabilistic extraction -- computing the prob...

Large language models don’t “learn”—they copy. And that could change everything for the tech industry.

Many unresolved legal questions over LLMs and copyright center on memorization: whether specific training data have been encoded in the model's weights during training, and whether those memorized dat...

Large language models don’t “learn”—they copy. And that could change everything for the tech industry.

YouTube video by AbbaVEVO

Plaintiffs and defendants in copyright lawsuits over generative AI often make sweeping, opposing claims about the extent to which large language models (LLMs) have memorized plaintiffs' protected expr...