Linus Tech Tips Discussion

 Insiders are encouraging mass "poisoning" of AI through the weaponization of information. Summary AI industry insiders have launched an initiative called "Poison Fountain" that encourages web developers to add "poisoned" code designed to destroy the cognitive integrity of LLM's trained on scraped web data. Their goal is to hinder AI progress and accelerate the collapse of the AI bubble by destroying the datasets AI companies rely on.           Quotes Quote "Alarmed by what companies are building with artificial intelligence models, a handful of industry insiders are calling for those opposed to the current state of affairs to undertake a mass data poisoning effort to undermine the technology." - Thomas Claburn, The Register. Quote "Poison Fountain was inspired by Anthropic's work on data poisoning, specifically a paper published last October that showed data poisoning attacks are more practical than previously believed because only a few malicious documents are required to degrade model quality." - Thomas Claburn, The Register. Quote "Our source said that the goal of the project is to make people aware of AI's Achilles' Heel – the ease with which models can be poisoned – and to encourage people to construct information weapons of their own." - Thomas Claburn, The Register.   Quote We agree with Geoffrey Hinton: machine intelligence is a threat to the human species. In response to this threat we want to inflict damage on machine intelligence systems. - RNSAFFN (Poison Fountain Website)   Quote "we found that as few as 250 malicious documents can produce a "backdoor" vulnerability in a large language model—regardless of model size or training data volume." - Anthropic   My thoughts At risk of huffing too much hopium in a drought of good news, I'm very happy to see a supposedly effective way to fight back against AI and the chaos it's causing. Though obviously for such a movement is to be successful, we rely on both enough of the right people becoming knowledgeable of it, and AI companies remaining unable to develop effective methods against it.      Sources RNSAFFN - Poison Fountain Website outlining their purpose, also provides links and instructions to their  "endless stream of poisoned training data". https://rnsaffn.com/poison3/   The Register - Original article reporting on the initiative  https://www.theregister.com/2026/01/11/industry_insiders_seek_to_poison/   Anthropic - Paper on the relative simplicity LLM poisoning https://www.anthropic.com/research/small-samples-poison Art Chad - The YouTube video that initially made me aware of this