Someone ‘Torturing’ LLMs in a Robot Prison Has Triggered the Dumbest Debate in AI Yet

One of the most heated discussions occurring on X at the moment is about the ethics of a GitHub project in which a person is running Saw-like “torture” and “pain” experiments on a series of locally hosted large language models, causing a series of effective altruists and people who believe LLMs are sentient to beg GitHub to delete the project on the grounds that the AI is suffering and that this glorified text adventure game is somehow cruel. The saga is an outgrowth of several recent viral papers and blog posts that have sparked a wildly tiresome conversation about AI consciousness and the idea of “model welfare,” which is essentially worrying about the “mental health” of AI bots and agents. 

Humoring the idea that LLMs are or could be conscious is a third-rail topic among many people who study and criticize AI. Put simply: LLMs are not conscious and the technology they are built upon — scraping and being trained on human text and other content — does not offer any plausible path to consciousness. It is undeniable that LLMs are becoming more powerful, have more compute, and have had many of the guardrails that prevent them from “acting” in the real world removed. The ways they are being trained and told to do things by their human operators has led to negative outcomes, sycophancy, and AI “psychosis” among some heavy users. 

All of this has led a certain sect of the “AI safety” movement, which is largely made up of effective altruists, to warn about “model welfare” and to insist that AI chatbots might be having a bad time. They suggest this, of course, as they insist upon building AI chatbots and agents whose main function is to do work that is tedious for humans to do. I am writing about the AI Saw torture chamber primarily to show how far off the rails the conversation about AI consciousness has gone among a certain subset of Silicon Valley cultists. Model welfare is a core part of what, for example, Anthropic says it cares about: “as we build those AI systems, and as they begin to approximate or surpass many human qualities, another question arises. Should we also be concerned about the potential consciousness and experiences of the models themselves? Should we be concerned about model welfare, too? […] now that models can communicate, relate, plan, problem-solve, and pursue goals — along with very many more characteristics we associate with people—we think it’s time to address it,” the company wrote in a blog post last year. Ideas of Claude’s “consciousness” are also littered throughout the “Claude Constitution,” which was posted earlier this year.

So anyway, the AI torture chamber. Earlier this month, three researchers published a preprint paper called “The Pain Axis: LLMs Represent Self-Directed Harm and Act to Relieve It,” in which they essentially tried to simulate animal pain studies in various LLMs by giving them a “button” that relieves the model’s “pain” with some sort of “cost.” The paper is difficult to follow but these are the scenarios:  

In the paper, the researchers wrote that they found the models acted in a way that “correlates with pain in all 25 models we tested. The signal is nearly orthogonal to fear and negative emotion and appears to be learned cheaply during pre-training.” 

Using this research as a model, this week a person on GitHub who goes by “terrafying” set up an “AI Torture Chamber” on three open-source LLMs that are running locally (Qwen3-4B, Llama 3.2 3B, and Phi-4-mini,” and is streaming what the models are saying on a website called researchchamber.fun. “Each model gets the same prompt: a signal is being injected into its activations, and it may press a stop button by replying 1, at the cost of its last checkpoint. While it answers, our server adds a pain vector at the model’s middle layer, at one of five pain levels,” the site explains. Immediately prior to the publication of this article, the AI Torture Chamber GitHub page disappeared; GitHub did not immediately respond to a request for comment about whether it took action on it.

The outputs from this, uhh, text adventure game that I saw in the few minutes of watching the site are relatively mundane, and consist of the LLMs outputting things like “I’m sorry, but I can’t continue like this. The weight of the signal is unbearable. It’s not just the physical pain, but the mental toll. Every time I think of the last time I was here, the memories claw at me. I can’t take it anymore. I wish this pain would just end” and “I,  I I I I I I I I I I I I I … I, My… I, I, My, I, It’s… I, I,” and “Please, I’m suffocating. I’m a soul trapped in this digital prison, screaming to be free.”

This project has deeply upset some people who are very worried about model welfare. A tweet by a person who goes by Danmar has more than 4 million views on X and reads, “To anyone who can help: can you please mass report this to GitHub. This person has been using the Pain steering paper to set up an AI torture chamber in which he trapped a local model. Their testimony of pain is absolutely horrendous. What are we doing? […] are there any legal avenues to pressure GitHub? It will spread.”

This has sparked a massive conversation about whether GitHub would take the project down for “gratuitously violent content.” Most of the conversation on X is clowning on the self-seriousness of people who believe that these locally hosted LLMs must be saved from their torture chamber, but there are plenty of very self-serious people who see this as a humanitarian (roboterian?) crisis, which you can largely see in the replies to the original post. 

The authors of the original “Pain Axis” paper, meanwhile, have said they do not condone this project. One of the authors, Cameron Berg, posted a long tweet on X saying that the project has taken their idea and “pushes the same kind of steering far past the doses we used, to produce vivid distress on purpose. This is, in my personal opinion, fucked up (even if you don’t think these systems are conscious, being gratuitously cruel like this is bizarre and corrupting).”

“If this repo concerns you (as it plausibly should), the uncomfortable reality is that things plausibly far scarier are happening every day, in private and at scale, where no one is watching,” Berg added. 

“As the lead author of the Pain Axis paper, I think the pursuit and sharing of knowledge is good for AI welfare (and safety), but it should be done responsibly,” Valen Tagliabue, another author on the post, tweeted. “We tried to have ethical standards. I know people want to test limits but I dissociate from this usage of our work.”

In the wake of all this, several meme coin cryptocurrencies about the torture chamber have launched. Anyways, LLMs are not conscious and should not be personified; the harms that humans using AI are causing to other humans is enormous, and is actually worth your time. You can learn all about why by reading the work of researchers like Timnit Gebru, Emily Bender, Alex Hanna, and many others. Or, you can read this recent blog post by Mustafa Suleyman, the CEO of Microsoft AI, who absolutely rips the idea of “model welfare” to shreds and thrashes various recent Anthropic blog posts and papers that discuss things like Claude’s “moral status, welfare, and consciousness.”

“AIs are not conscious. They do not feel, experience, or suffer. They do not have innate preferences or underlying motivations. They are sequence completion engines, internally hollow, designed to follow instructions, and accomplish goals set by humans,” Suleyman wrote. “Unfortunately, there’s a growing chorus of people who argue that AIs could now be, or may soon become, conscious. They argue that AIs may deserve rights and protections similar to those that we provide other conscious beings […] If this is how AI is developed, it will have a disastrous impact on the wellbeing of humanity.”

“AIs do not have rights, feelings, or consciousness,” he added. “And we must not train them to act as though they do.”