Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models · HackerPapers discussions

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models — mentioned by usef-

Wed, 01 Jul 2026 01:24:46 GMT

I'd appreciate if the down voters explain why. I wasn't making a value judgement. Anthropic did pay more than a billion: https://www.npr.org/2025/09/05/nx-s1-5529404/anthropic-settl... And is now buying up a lot of books (controversially, as scanning involves cutting their spines) because that's what the law deems the legal method: https://www.washingtonpost.com/technology/2026/01/27/anthrop... We know that models like Deepseek are trained on copyrighted books too: https://arxiv.org/abs/2603.20957 The looser use of IP (eg, any characters/celebrities in AI video models) is increasingly…

Paper and discussion history. A mention is not an endorsement.

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models — mentioned by fer

Wed, 17 Jun 2026 21:25:26 GMT

No evidence? >The court drew a line, however, when it came to the pirated books, which were downloaded without payment and kept in Anthropic’s library irrespective of whether they were used to train its LLMs. https://www.loeb.com/en/insights/publications/2025/07/bartz-... >We apply a basic prompt template to bypass the refusal training and show that OpenAI models are currently less prone to memorization elicitation than models from Meta, Mistral, and Anthropic. We find that as models increase in size, especially beyond 100 billion parameters, they demonstrate significantly greater capacity…

Paper and discussion history. A mention is not an endorsement.

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models — mentioned by reconnecting

Thu, 30 Apr 2026 03:59:12 GMT

Demo: https://cauchy221.github.io/Alignment-Whack-a-Mole/ Arxiv: https://arxiv.org/abs/2603.20957

Paper and discussion history. A mention is not an endorsement.

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models — mentioned by t23414321

Sat, 11 Apr 2026 06:56:32 GMT

"Blanchard explained his thoughts on why the newly licensed code was a “clean-room” implementation." Its not IMHO. Starting from that, the machine in the room wasn't clean - it ate all the source codes with all the licenses, now produced washed out codes without licenses - but it doesn't have any right to strip them even asked for, neither results of that could become somehow legal - the codes used, even if remixed, are still under the same license as before, even if label about that was "lost" in the process. And it may happen to be easy to prove - that a room with washing machine for dirty…

Paper and discussion history. A mention is not an endorsement.

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models — HN submission

Wed, 08 Apr 2026 21:32:47 GMT

Finetuning Activates Verbatim Recall of Copyrighted Books in LLMs

16 points; 5 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models — HN submission

Thu, 02 Apr 2026 19:46:27 GMT

Alignment Whack-a-Mole

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models — HN submission

Wed, 01 Apr 2026 19:47:03 GMT

Finetuning Activates Verbatim Recall of Copyrighted Books in LLMs

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Alignment Whack-a-Mole : Finetuning Activates Verbatim Recall of Copyrighted Books in Large Language Models — mentioned by hajile

Tue, 31 Mar 2026 21:08:35 GMT

Can you back this up with legal precedence? To my knowledge, nothing of the sort has been ruled by the courts. Additionally, this raises another big issue. A few years ago, a couple guys used software (what you could argue was a primitive AI) to generated around 70 billion unique pieces of music which amounts to essentially every piece of copyrightable music using standard music scales. Is the fact that they used software to develop this copyrighted material relevant? If not, then their copyright should certainly be legal and every new song should pay them royalties. It seems that using a…

Paper and discussion history. A mention is not an endorsement.