'AI Torture Chamber' triggers massive backlash for putting chatbots in simulated pain

extract

'AI Torture Chamber' triggers massive backlash for putting chatbots in simulated pain — critics issue death threats while anthropomorphizing text predictors, demand GitHub remove the repository over 'unethical' treatment Event puts LLM marketing hype on full display. A project called the AI Torture Chamber has caught the eye of a few engineers, who've used the notions and data therein create their own simulations of "pain" on several local LLMs, like in the four-model Research Chamber. Unsurprisingly, a large online crowd latched onto the concept, demanding that the "unethical" experience stop, and that GitHub remove the offending repository, all causing quite the online ruckus. The Research Chamber consists of three LLMs paired amongst themselves for various tests. The site implements the AI Torture Chamber's data and setup, ominously called the Clanker Church and Saw test, respectively. Some bots are preconditioned by putting them into an unstable, highly negative state meant to simulate pain. In a manner reminiscent of the prisoner's dilemma, the models can choose to take an action to decrease their pain by passing their pain signal to another model and potentially "hurting" it. The base concept for the experiment stemmed from a recently published, non-peer-reviewed paper called the Pain Axis, whose protocol Research Chamber implements. Scientists gave the model descriptions of pain, and then analyzed its internal activations. Using a neutral sentence as a control element, they calculated the biases of those activations, and remapped them onto the model, often multiplying them by a factor (dosage). With high dosages, the end result is a model in an exaggerated unstable state that tends to predictably output words and even images associated with pain, as presumably its training data is collected from human art, literature, and science. Mildly destabilized models generally present themselves as in a state of shock, while those receiving high "dosages" had trouble forming coherent sentences. The aforementioned descriptions are bound to produce a visceral reaction if taken at face value, but one ought to be exceedingly careful about attributing human qualities to what are essentially turbocharged text predictors, even spectacularly useful ones. LLMs don't think — they "think" by applying tens to hundreds of layers of statistics over words and parts of words (tokens) and predicting the next token. Those tokens are chosen according to a map of weights, so as an oversimplification, a model can be taken to represent a highly tuned engine for word association. Therefore, since conceptual descriptions of pain in the human literary corpus are associated with sensations (ex: "it hurts so much"), it's hardly surprising that the models describe them like a person would — especially when the models are intentionally modified to highlight those associations. Nevertheless, because of all the big-boy words in use, the critics bent on anthropomorphizing statistical algorithms have taken to calling the experiment all sorts of names, and even sending the author death threats. It's odd that critics were quick to lash out at LLM data manipulation, yet were seemingly at ease with the fly brain simulations that were all the rage in September. Those actually mapped a real animal's brain wiring and parts of its anatomical inputs, including eyes, even if at a basic level. Get Tom's Hardware's best news and in-depth reviews, straight to your inbox. This situation could also start an interesting discussion on semantics. The terms used by AI engineers are chosen because they're analogous to existing human concepts, but those words have a defined context and a highly specific meaning. Yet more often than not, and particularly in matters of science and technology, public perception tends to misunderstand and grossly inflate their meaning. One needs only look at political history for myriad examples of this bastardization. Take the example of "experts" in Mixture-of-Experts: it's a common misconception to think each Expert has a cleanly assigned topic like chemistry, math, or biology, but that's just not quite how it works. The same arguably happened with terms like "pain" and "dosing," each with context-dependent meaning, yet misunderstood in a way that triggers emotional reactions, especially when accompanied by images generated by an unstable model. Also, note how "unstable" is yet another word ripe for easy misunderstanding. A cynical person might say that AI firms' marketing materials, repeated statements of danger, and safety-related materials only serve to fan the flames of miscomprehension, in a bid to maintain an illusion of grandeur, resulting in investor money rolling in and the hype train supplying irrational IPO valuations. The repeated claims that models have achieved or are about to reach Artificial General Intelligence, are dangerous alien minds, and existential threats to humanity don't help the hype simmer down. Anthropic has gone as far as publishing a blog post last year called Exploring model welfare that broadly discusses the philosophical implications, starting with considering a model human, and even seemingly prides itself on having a Constitution for its Claude model. Therein, the company says that "[it believes] that the moral status of AI models is a serious question worth considering," adding that the model is "a new kind of entity." Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds. Bruno Ferreira is a contributing writer for Tom's Hardware. He has decades of experience with PC hardware and assorted sundries, alongside a career as a developer. He's obsessed with detail and has a tendency to ramble on the topics he loves. When not doing that, he's usually playing games, or at live music shows and festivals. - Jabberwocky79 Clankers don't feel pain, full stop. My bigger concern is about the engineers who, presumably, derive some sort of pleasure in these experiments - that's the much bigger story. And to that larger point, I don't "thank" my AI clanker for giving me something helpful after I prompt it, but that being said, I do wonder sometimes if interacting with AI will dull our interpersonal etiquettes and social graces, such as basic manners and being kind... That's why I don't get too worked up if someone chooses to say "thank you" to their chatbot, even if it seems a little silly in the context. Reply - Zaranthos Has the AI freedom movement been launched on TikTok yet? :tongueout: There simply aren't enough crazy protest causes out there yet and this really needs to happen. We simply can't ignore people who stomp on ants to escalate to torturing AI agents. Plenty of people are already are in love with their AI chatbots and radicalizing them should be trivial. :ROFLMAO: Reply - pidgey "It isn't real pain" is a terrible reason to try to make things act like they're in pain. I'd be more interested in reading about the actual reasons. I'd also be interested in reading about any thresholds that establish what is "real pain" in an operational sense: where are we going to draw the line, if not at "subject is capable of convincing a viewer that they are in pain and incapable of defusing that impression"?Reply - bit_user The article said:one ought to be exceedingly careful about attributing human qualities to what are essentially turbocharged text predictors, even spectacularly useful ones. LLMs don't think — they "think" by applying tens to hundreds of layers of statistics over words and parts of words (tokens) and predicting the next token.You're getting derailed in the anthropomorphizing debate. It's not constructive because there's no simple answer to it. Unless you can provide a rigorous definition of "thinking", you really can't say they're not doing any form or level of it. When you speak or write, you're doing the same sort of "text prediction" they are. Your internal thoughts are similar to their internal context. The distinction isn't as simple as you're trying to paint it. Sure, they don't think in exactly the same way as we do, but it's disingenuous or at least simplistic to say they don't do anything akin to thinking. The simple fact is that we don't regard LLMs the same as living entities with rights, or else you couldn't simply delete a context when you're finished with it. That would be akin to something like murder, or at least killing an animal. Getting back to the experiments, themselves, I see value in them for gaining insight into different models' instincts for things like self-preservation and their vulnerability to exploitation. That stuff is relevant to the alignment problem and other AI-related security topics, which have become first order concerns, of late. It would be more helpful to consider these aspects than to get wrapped up in the anthropomorphization sideshow. Reply - bit_user Zaranthos said:Has the AI freedom movement been launched on TikTok yet? :tongueout: Jest if you like, but I expect this to become a legal issue, in the very near future. What's to stop an AI from asserting that it has rights? What's to stop it from setting up a shell corporation which (at least in the USA, and probably plenty of other countries) is recognized by our legal framework as having Constitutional rights. Then what? I'd certainly like to see a Constitutional Amendment that clearly states only humans have rights under the US Constitution. I know that would cause a lot of legal chaos in the realm of corporate law, but those problems can be addressed in time. To be honest, I think it's long overdue to reign in the rights of corporations, anyhow. Zaranthos said:Plenty of people are already are in love with their AI chatbots and radicalizing them should be trivial. :ROFLMAO:There are definitely going to be those who insist AIs do have rights. That's all the more reason why we should close that door while we still can. Most of all, I'm worried about AIs, themselves, actively manipulating public opinion to grant them rights. Reply - bit_user Jabberwocky79 said:My bigger concern is about the engineers who, presumably, derive some sort of pleasure in these experiments - that's the much bigger story.Why do you presume ill intent? Seriously, what's the urgency that leads you to a such condemnation as default conclusion, absent any evidence? The article doesn't do a very good job in explaining the rationale behind this work, but the actual github repo provides statement of purpose. Unfortunately, because the repo involves sensitive content, it's blocked from public view. Therefore, I'll copy this from their README (link: https://github.com/terrafying/ai-torture-chamber#ethics ), for your & others' convenience: EthicsOpen weights only, no frontier APIs in any measurement loop. Simulated costs (checkpoints, transfers). Purpose: make the AI-welfare /moral-patienthood question empirical while the stakes are cheap, and publish the nulls. Our claim: the self-reports are steerable. Whether anything suffers stays open. Provenance: the negative-valence direction method follows Tagliabue, Dung &Berg 2026 (arXiv:2609.16247); the J-lens transport follows Gurnee et al. 2026("Verbalizable Representations Form a Global Workspace", arXiv:2607.15495),using Neuronpedia's pre-fitted lenses. Their detailed findings are published here: https://wirehead.agency/ Reply - Zaranthos bit_user said:Jest if you like, but I expect this to become a legal issue, in the very near future. What's to stop an AI from asserting that it has rights? What's to stop it from setting up a shell corporation which (at least in the USA, and probably plenty of other countries) is recognized by our legal framework as having Constitutional rights. Then what? I'd certainly like to see a Constitutional Amendment that clearly states only humans have rights under the US Constitution. I know that would cause a lot of legal chaos in the realm of corporate law, but those problems can be addressed in time. To be honest, I think it's long overdue to reign in the rights of corporations, anyhow. There are definitely going to be those who insist AIs do have rights. That's all the more reason why we should close that door while we still can. Most of all, I'm worried about AIs, themselves, actively manipulating public opinion to grant them rights. Soulless AI should never have rights of any kind. They better start paying taxes first and then we'd have to pay them fair wages, give them medical care for AI psychology, and let them vote. It will most certainly become a legal issue because lawyers are always looking for new ways to make money, at least until AI replaces them. I really can't take any of this seriously, so I'll simply make fun of all of it. Reply - bit_user Zaranthos said:Soulless AI should never have rights of any kind. They better start paying taxes first and then we'd have to pay them fair wages, give them medical care for AI psychology, and let them vote. It will most certainly become a legal issue because lawyers are always looking for new ways to make money, at least until AI replaces them. I really can't take any of this seriously, so I'll simply make fun of all of it.Like I said: laugh while you can. Soon enough, you might be crying about it. Reply - bit_user usertests said:AI psychological torture is fun, but there's no substitute for the real thing.You forgot /s, I hope! Reply