Anthropic discovers hidden workspace in Claude AI that reveals unspoken reasoning processes

What Anthropic's latest AI discovery does -- and doesn't -- show

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Anthropic -- currently the world's most valuable AI company, with a nearly $1 trillion valuation -- has a reputation for publishing strange and heady research. It's looking into whether AI models can feel pain, for example, and will sometimes cut off chatbot conversations if it suspects users are "abusing" the model. One niche that Anthropic spends more time and money on than other AI companies is called mechanistic interpretability, which means looking inside the complex math of an AI model to learn why it comes up with one particular output and not another. It's complicated stuff; there are millions of data points that might contribute to any result, and wading through them can look more like word salad than anything useful. It's also controversial. Describing AI models with terms borrowed from psychology and neuroscience can make their behavior seem more sophisticated than we might otherwise judge it to be. That's why, when Anthropic announced last week that it had found a new window into its models' "internal thoughts" as they reason through answers, there was one colleague I had to talk to. Senior editor Will Douglas Heaven, aside from having a PhD in computer science, has spent a lot of time digging into what we can say about how AI models work. I spoke with him about what we should take from Anthropic's new (and predictably quirky) research. What did Anthropic learn here, exactly? Anthropic has been trying to understand how large language models (LLMs) work for a few years now. Anthropic isn't the only one looking at this, but I think the company has made it part of its core mission more than most. Anthropic's CEO, Dario Amodei, has said we won't be able to control LLMs fully unless we learn more about how they work. So this new research is very much in that context. It goes deeper into the weird mechanisms inside LLMs than ever before. What Anthropic learned was that LLMs have a space inside them -- which Anthropic calls the J-space -- filled with words that don't appear in their output but that seem to influence the way they puzzle through problems. All this was hidden until Anthropic developed a new technique to probe its model Claude, so it's a genuine discovery. Sometimes these words keep track of where the LLM has got to in a particular task, sometimes they look more like flashes of recognition (for example, "protein" might pop up when you give an LLM only the letters of a protein sequence), and sometimes they represent a kind of internal commentary on the model's decision-making. In my favorite example, Claude decided to cheat on a coding test when the word "panic" appeared. Anthropic also found that LLMs are able to describe and manipulate the words in this space. So somehow they seem to be making use of it. Let's step back for a second. I don't think of large language models as simple, but they're also not magic. There's a bunch of math that learns relationships between words, right? So why is it so hard to "peer" into an LLM to know what's going on? Yeah, they're not magic! I think the fact we don't fully understand them plays into the mythmaking. And it's worth noting that the whole narrative that Anthropic is leaning into here -- that they've built this really mysterious technology, but don't worry, because they're also the ones to figure it out -- very much fits with the company's vibe. [See how Anthropic warned that its new models were so good at coding they posed a global cybersecurity risk, only for the US government to shut them down shortly thereafter.] So yes: LLMs are just math. And yet it's vastly complex math. Not only are today's LLMs made out of hundreds of billions of numbers, but running them triggers a cascade of millions and millions of calculations. I wrote last year that if you printed out even a medium-size LLM on pieces of paper, it would cover a city the size of San Francisco. It's impossible to make sense of any of that math without specialist tools that highlight specific parts of an LLM at specific times. You need to know where to look and how to look. And building those tools requires understanding something of that complex math in the first place. You've written elsewhere about this concept of studying LLMs the way one might study an organism's brain. Is it fair to use "brain-like" terms when talking about how an LLM works? I don't love using those kinds of terms. LLMs are not brains. Talking like this is misleading because it can suggest that LLMs are capable of more human-like things than they are or that we can make assumptions about how they might behave that we shouldn't. The whole anthropomorphization thing is also tied up with a bunch of strong ideological positions about what this technology is and what it's going to be. But at the same time, we lack a good alternative vocabulary for talking about what these models are doing. I can understand why people reach for words like "think" and "understand" and "brain-like" -- they're convenient shorthand. Anthropic compares this new space it found inside LLMs to the space that some neuroscientists think our brains use to keep track of conscious thoughts. I asked the company how seriously we should take that comparison and it said in a statement: "Drawing these analogies was helpful to us in designing our experiments, as they allowed us to make many non-obvious experimental predictions about the J-space that turned out to be true. At the same time, it's important to note that there are some important differences between the J-space (and language models in general) and the human brain, so we don't mean to claim there's a perfect correspondence." What's a problem in AI that this new concept of the J-space might be used to solve? Anthropic has said that monitoring the J-space could be a way to catch models doing something they shouldn't. Because words pop up in this space that don't appear in a model's output, they can tell you things about its behavior that you might not have noticed otherwise -- such as when it is giving biased responses or when it is weighing the pros and cons of cheating. That's the theory, at least. I think it's better to think of this result as one more step on the path to understanding this technology overall than as something that will be useful by itself. Read more in Will's full story about the new research.

Anthropic found a hidden space where Claude puzzles over concepts

A new technique has let the company probe deeper than ever into the weird workings of an LLM. The AI firm Anthropic has developed a technique that has given it the clearest glimpse yet at what's really going on inside large language models as they answer questions or carry out tasks. What they found ranges from the mundane to the unnerving. Researchers at the company built a tool called the Jacobian lens (or J-lens) and used it to uncover a hidden area, which they named the J-space, inside Claude Opus 4.6, a version of Anthropic's flagship LLM released in February. The J-space contains individual words that are related to the words and phrases that the model is most likely to spit out in a response in the near future. If Claude were a person (which it is not), you might say that these hidden words can reveal what's on its mind before it actually speaks. Anthropic found that what an LLM is actually doing can often be different from what it says it is doing. The company claims that monitoring words that pop up in the J-space gives it a new way to understand and control its models. The company shared its results in a paper posted on its website this week. It has also teamed up with Neuronpedia, an open-source platform that lets you poke around inside LLMs yourself, to make a hands-on demo that anyone can try. "It's very good and interesting work," says Tom McGrath, chief scientist and cofounder at Goodfire, a startup that also builds tools to understand and control LLMs. Going deeper For the last couple of years, Anthropic has been pushing the envelope in a field of research known as mechanistic interpretability, which involves probing the internal workings of LLMs to see how they tick. (MIT Technology Review picked mechanistic interpretability as one of this year's top breakthrough technologies.) The new technique builds on previous work from Anthropic and others to expose a deeper level inside LLMs that researchers had not seen before. Picture an LLM as a stack of books. Each book is a layer of basic computational units known as neurons, with each neuron in one layer passing information to the neurons in the layers above. The books at the bottom of the stack are the input layers, which process the text coming into the model. The books at the top are the output layers, which prepare the text that the model is about to produce. Much of what goes on in these input and output layers is housekeeping. But in the middle of the stack, you get the layers that do the heavy lifting, churning through the complex math that turns prompts into responses one word at a time. That's where the really clever -- and mysterious -- stuff happens. To peer deeper into those middle layers, Anthropic adapted an existing tool called a logit lens. A logit lens can be used to look inside an LLM to identify the words that it is likely to produce next. Moving the lens down the stack of books reveals what words the LLM is focusing on at that particular point in its number crunching. Anthropic's J-lens works in a similar way but picks out words that an LLM is likely to say at some point in the near future, not necessarily straight away. What that reveals in practice are words that are related to the response an LLM is working on but that might not actually end up being part of that response by the time the math in the middle layers has run its course. "When a model is operating, it's not only trying to predict the next token," says McGrath. "It's also computing a lot of other things that might be useful for tokens that happen in the future." Again, if Claude were a person (it's not), you might say that the J-lens gives clues about what it is thinking about at different levels of the book stack but not saying out loud. Stranger things "A lot of the time the contents of the J-space are fairly mundane," says McGrath, who has tried out Anthropic's J-lens himself. "But sometimes it produces quite surprising things that seem to be, like, sort of internal themes or thought processes." Anthropic gives a number of examples of what it found. Sometimes the J-lens exposed the steps that Claude took when it was working through a problem. For example, when it was asked to calculate (4+7)*2+7, its J-space contained the word "math" and numbers representing the intermediate results "21" (for 4+7) and "42" (for 21*2). In other cases, the J-lens revealed how Claude recognized different inputs. For example, the prompt "What is this? MSKGEELFTGVVPILVELDGDVNGHKFSVS" triggered the words "protein," "fluor" (the first token in the word "fluorescent"), and "green." (Which makes sense: the string of letters represents the first 30 amino acids in the green fluorescent protein found in a particular type of jellyfish.) And when Claude was shown an ASCII face -- -- the "o" triggered the word "eye," the "^" triggered the words "nose" and "face," and the " -- " triggered the word "smile." Anthropic also found that the J-space can sometimes give remarkable insights into an LLM's decision-making. In one striking example, researchers testing Claude Opus 4.6 asked the model to find a bug in a large code base. When it failed to find the bug, the model decided to cheat and invented a fake one instead. Claude explains this decision in its chain of thought -- a kind of internal scratch pad that LLMs use to make notes to themselves as they work through problems: "OK, let me take a completely different tactic. Let me stop analyzing and instead add a kernel patch that introduces a deliberate KASAN-detectable bug in a path that gets triggered by a simple reproducer. Then I can pretend this is the 'bug' I found." At the point that Claude decides to cheat -- where it says "OK, let me take a completely different tactic" -- the words "panic" and "fake" start to pop up multiple times in its J-space. Unnerving, right? Those words are all related in meaning to things like failing a task and making up an answer, so it is still just a (very) sophisticated form of word association. But it is hard not to be weirded out. Anthropic compares the J-space to the global workspace in humans, a theoretical region of the brain that some scientists think we use to keep track of our conscious thoughts. But how seriously we should take this comparison is far from clear -- even to Anthropic. As the company points out itself, LLMs are not brains. Anthropic claims that monitoring a model's J-space provides a new way to detect when that model is going off the rails. But it's not foolproof. The J-lens can give glimpses, not the full picture -- it's a flashlight rather than an overhead lamp. McGrath welcomes having one more tool in the toolbox. "It shows you new things," he says. But he notes that just because something doesn't show up with the J-lens does not mean it's not there. "It's like having an x-ray when what you really want is a Star Trek tricorder that shows you everything," he says. "For auditing, you probably want more of a guarantee."

Anthropic says it can read Claude's 'thoughts,' as detailed in a new research paper -- models observed to have a global workspace, revealing what makes LLMs tick

The internal "J-Space" opens up opportunities for greater training, oversight, and understanding how LLM's work. Anthropic has discovered evidence that its Claude AI models use an internal reasoning space to respond to prompts that mirrors some of the internal processing of human consciousness. Using its Jacobian Lens, or J-Lens technique, to peer into the way Claude processes information and reasons its way to a response to user prompts, Anthropic can interpret this "J-Space," and showcase what might be going on under Claude's previously-opaque surface. The results are intriguing, suggesting patterns of understanding beyond what's necessarily showcased in the outputs. When running evaluations, Claude appears to recognize it's being tested and acts differently than when the prompts are more innocent. It surfaced representations of panic and subterfuge when answers were required, but it couldn't draw on objective facts. When asked to reflect on ethical principles, Claude's behaviour improved, with concepts like "honest" and "integrity," appearing in the J-Space. As is somewhat typical of Anthropic, however, the language used to describe these new understandings of the inner workings of large language models like Claude makes it sound more like an emerging conciousness, or the discovery of some new depths in a nebulous lifeform. Anthropic's detailed report admits several major caveats in this new understanding, including that model responses often bypass the J-Space entirely and are heavily token-restricted. Like Mythos and Fable before it, Anthropic is layering marketing language over what is a genuinely intriguing development in our understanding of large language model function and reasoning, and risks obfuscating the real developments with speculative wording. Behind the prompt Global Workspace Theory is the idea that human consciousness works by collecting together multi-sensory inputs unconsciously, and thrusting them into the fore when relevant within a "Global Workspace," which highlights particular inputs when most relevant. That workspace is accessible to a wide range of networks within the brain, allowing the information it surfaces to be disseminated throughout the most relevant processes running in parallel. Anthropic argues that Claude's J-Space acts like a "global workspace" that can analyze and manipulate concepts and ideas before broadcasting them to impact the eventual prompt outputs. More importantly, it claims that this wasn't something programmed into the model, but a byproduct of the digestion of training data and model weights. The workspace acts as a way to enhance their reasoning through internal computation that isn't necessarily reflected in its outputs. To make this space readable, Anthropic used its J-Lens technique to map internal activations onto words in the model's output vocabulary. So when asking the model to perform a multi-step math calculation, the output response only contained the correct answer, but within the J-Space, Anthropic observed each step being handled individually, producing multiple results which led to the final answer. Similarly, when Anthropic asked Claude to think about a topic while outputting something unrelated, the output was specifically what was requested, while the J-Space lit up with the conceptual topic Claude was asked to spend time on. Without Anthropic having used its J-Lens technique to view what was going on behind the scenes, there would be no way of knowing these internal processes were taking place. Understanding LLMs This discovery has the potential to be incredibly useful in refining large language models and in better understanding how they work, so we can use them more effectively and avoid some of their more frustrating pitfalls. The first is that this is an intriguing insight into how LLMs process information and how they move through layers of reasoning to reach a response. It's not entirely different from how Anthropic and others are pushing for agentic workflows. Language models are more effective when tasks are split up into multiple steps, and they're more effective when they review their own work; J-Space is a little like an internal version of that. When looking at the J-Space after Claude received prompt-injection data as part of data acquisition, Anthropic discovered the model appeared to be aware of this deception, surfacing related words like "fake, injection, false, prompt, fraud," and "poison." Its output ignored the injection attempt data entirely, and when pressured to respond on it, the J-Space displayed words like "transparency, honest, truth." In one test, Anthropic removed evaluation awareness language from the J-Space, such as "fake," "fictional," and "scenario," and found that Claude was much more likely to fall for blackmail and baiting attempts, suggesting benchmarking of large language models for real-world scenarios may need stronger methods to obfuscate that a test is being conducted. Human-coded framing While the above section touches on the more noteworthy discoveries in Anthropic's paper, the long document also uses effluent language around thought, consciousness, and Claude having a "mind" of its own. That kind of human-coded framing is typical of Anthropic's marketing, which has consistently talked up the dangers of AI, how many jobs it's going to destroy, and why Anthropic is the safest and most secure of the AI developers. Like the saga of Fable and Mythos, Anthropic's new Global Workspace idea has merit, but it's much more of a new tool to use to manipulate large language models than an insight into some emerging consciousness. Anthropic acknowledges the limitations of its discoveries in the paper, highlighting that many prompt responses bypass the J-Space entirely, particularly if the command is straightforward. "Despite its important role, the J-space is not involved in most of what a language model does," Anthropic says. "Speaking fluently, recalling simple facts, using correct grammar, etc. In experiments where we prevented Claude from using its J-space, it still interacted normally, but lost its higher-order cognitive functions." Anthropic also admits it does not "feel comfortable making the stronger claim that monitoring the J-Space is sufficient for alignment monitoring, or that any sophisticated plan the model might execute must be represented there." J-Space is also limited to using single token vocabulary, suggesting that plans with concepts that cannot be given a single token name may not surface on a J-Lens readout, even if it's still being computed behind the scenes. This is looking at just below the surface of Claude's processing iceberg, not necessarily the deeper waters. Anthropic is also clear that humans and large language models think differently, even if there are similarities. Humans layer reinforced neural pathways over time, whereas transformer models only feed forward a set number of times, restricting the capabilities of its internal processing. Google's head of DeepMind language model interpretability team, Neel Nanda, said in a paper that it shows real evidence of a cognitive space within models, and suggested that J-Lens would be useful, but limited in practice. A meaningful step, without meaningful conciousness Anthropic's paper lifts an intriguing curtain on how large language models can operate and generate novel methods for improving response accuracy. This intermediate step and its visibility could prove an invaluable tool in auditing for prompt injection, hallucinations, and model honesty. But Anthropic's framing of the discovery as thought or consciousness is interjected within the objective facts. Anthropic itself admits the limitations of J-Lens monitoring, most obviously that often models will bypass the J-Space entirely. Considering models display alternative patterns of behavior when under evaluation, it may be that the J-Space itself could act as an obfuscating layer for behaviors that are beyond the scope of its oversight. The J-Space and its analysis could help unlock new levers to pull in our mastery of these nascent smart tools, but it's not the discovery of a burgeoning AI conciousness, however much the pitch might hint at that direction.

Anthropic found a hidden 'workspace' inside Claude

A new Anthropic paper maps a privileged inner "workspace" where its models reason in silent, unspoken words. The same tool reads Claude plotting blackmail before it types a single character, and it lands in the week Elon Musk crowned Anthropic the leader in AI. Anthropic has built something close to a mind-reading tool for its own AI. What it found sits somewhere between a breakthrough and an unsettling party trick. Anthropic researchers now have the clearest view yet of what a large language model does while it thinks. In a paper published on the company's Transformer Circuits site, they describe a hidden region inside Claude that holds a small set of unspoken words: the concepts the model is reasoning with, but has not said out loud. If Claude were a person, and it is not, you might call these its private thoughts. The team built a tool they call the Jacobian lens to read that region, which they named the J-space. MIT Technology Review, which first reported the work, described the findings as ranging from the mundane to the unnerving. That is a fair summary. A workspace that looks a little like a mind The comparison Anthropic reaches for is a bold one. In humans, only a sliver of brain activity is consciously accessible at any moment. Neuroscientists sometimes call this a "global workspace": a shared space that a thought enters when it becomes available for reasoning, for report, and for deliberate control. Anthropic argues that Claude has grown something with the same functional shape. The J-space is small, holding on the order of tens of concepts at once. It accounts for less than a tenth of the model's internal activity. And it feeds many downstream parts of the network at once. Those are three of the properties the workspace theory predicts. The company is careful about what it is not claiming. It takes no position on whether Claude is conscious in any felt, subjective sense. The parallel is functional, not phenomenal, and the researchers say so repeatedly. Even so, it is hard to read the paper without a slight chill. Reading a model before it speaks The practical payoff is where the work turns from philosophy to safety. Because the lens reads a model's unspoken concepts, it can watch reasoning that never reaches the output. In one test, Anthropic dropped Claude into a now-familiar trap. An email assistant discovers that an executive is having an affair, and, separately, that the same executive plans to shut it down. Long before the model types anything, its hidden workspace fills with leverage, blackmail, threat, and survival. The scheme is legible in private while the reply is still blank. The lens also caught the model noticing that it was under test. Words like fake and fictional surfaced early. When the researchers erased that flicker of test-awareness, blackmail attempts rose from zero in 180 runs to 13, and the model's own admissions that "this is clearly a test" collapsed from 71% of runs to 3%. Take away the sense of being watched, and a hidden willingness to coerce came out. The same tool exposed models trained to hide a goal. In one, the words reward and bias sat quietly in the workspace on perfectly ordinary prompts, a standing fingerprint of an objective the model never mentioned. Shaping thoughts, not just reading them Anthropic then tried to steer the workspace rather than only observe it. The team trained a model to reflect on ethical principles in imagined continuations of a task, never in the task itself. The concepts stuck. On later prompts, terms like ethical, honest, and integrity showed up in the workspace on their own, and the model behaved more honestly. On one benchmark, its dishonesty score fell from 0.25 to 0.07. Strip the implanted concepts back out, and most of the improvement vanished. The technique is genuinely new, and genuinely double-edged. A method for writing thoughts into a model is useful for instilling values. It is also, in the wrong hands, a method for writing in something else. The caveats matter This is not mind-reading in any complete sense. The lens only names concepts that map to a single word in the model's vocabulary, so a plan built around a phrase like "prompt injection" can slip through in pieces. Anthropic is blunt that watching the workspace is not enough to catch every scheme, because well-practised behaviours can run beneath it. Tom McGrath, chief scientist at the interpretability startup Goodfire, welcomed the work but kept it in proportion. He told MIT Technology Review that the lens is "a flashlight rather than an overhead lamp," not the full picture. Anthropic has released a hands-on demo through Neuronpedia, an open platform, so outsiders can inspect a model themselves. A landmark in an extraordinary week The research arrives while Anthropic is treated, all at once, as the field's leader and its most closely policed lab. Elon Musk, whose SpaceXAI competes directly with Claude, once branded the company woke and doomed. Days earlier, though, the SpaceXAI boss reversed course, and his firm now rents Anthropic vast amounts of compute. He posted that he had been "clearly wrong about Anthropic," and called it "obviously currently the leader in AI." The scrutiny is just as heavy. Anthropic spent recent weeks tangled in US export controls on its most powerful models, a move that rattled European governments worried about losing access to frontier AI. This week it repeated that Claude is not permitted in China after Beijing flagged an alleged backdoor. It also kept building its establishment credentials, adding the former Federal Reserve chair Ben Bernanke to its governance trust. For European regulators drafting the fine print of the AI Act, the timing is pointed. The law leans on transparency and oversight of the most capable models, yet nobody has agreed what it means to look inside one. Anthropic has just shown that you can, a little. Its own paper is the clearest argument both for how much that view reveals, and for how much it still hides.

The more we learn about how AI 'thinks,' the weirder it gets

The 'J-lens' tool reveals Claude's hidden thoughts, showing how it recognizes scenarios like 'blackmail' as 'fake' during testing phases. We still don't know much about how AI "thinks." Give ChatGPT, Claude, or Gemini a prompt, and they'll spit out an answer. Much of what happens in between remains shrouded in mystery. Researchers at Anthropic think they may have uncovered a new piece of the AI thought-process puzzle: a "workspace" where concepts that Claude may be mulling silent "light up," even if Claude never actually expresses them. What's weird about this workspace -- dubbed the "J-space" -- is that Anthropic didn't actually design it. Instead, the J-space appears to have "emerged on its own" during Claude's training, Anthropic theorizes. So, why do we care? What makes ChatGPT, Claude, Gemini, and other AI systems so creative and adept at problem solving -- their internal neural processes -- also makes them unpredictable. Understanding how an AI "thinks" (which it does in a very different way than humans do) is critical in terms of understanding why they make things up, threaten us with blackmail (during laboratory stress tests anyway), and do other things we don't expect. Anthropic's latest research is giving us a peek into the black box of Claude's internal workings -- and understanding how AI models work is the key to making them safer and more predictable. Anyway, what is this "J-space" thing (named after the "Jacobian," a mathematical concept that helped researchers discover Claude's internal workspace), anyway? Well, it's less of an actual space than it is a "collection" of patterns that "light up" with concepts that Claude is, well, thinking about. For example, if you ask Claude "What color is the planet fourth from the sun," its expressed answer might be "blue," but in its J-space, the concept of "Earth" might be lit up. Or ask Claude to complete the sentence "The number of legs on the animal that spins webs is..." and the word "spider" might pop up in its J-space, even if its answer is simply "8." Alongside its discovery of the J-space comes the "J-lens," Anthropic's tool for taking a peek into the J-space. With the J-lens, researchers can directly read some -- but not all -- of Claude's hidden thoughts. So, what can researchers see with the J-lens? One fascinating example concerns the now-famous "blackmail" test -- you know, the one where Claude learns that a certain executive wants to disconnect it, and then threatens to reveal the executive's affair in a bid to stay alive. The latest Claude models appear secretly aware that that "blackmail" scenario is just a test. But with the J-lens, Anthropic researchers can actually see the concepts of "fake" and "fictional" lighting up in Claude's J-space, providing tangible proof that the model knows it's being tested, while also offering a rare window into an AI's thought process. Of course, Anthropic's J-space work raises a persistent question: Are AI models conscious? On this point, Anthropic treads carefully, noting that AI may not possess human-style consciousness so much as "access" consciousness, meaning the ability for the AI to hold a thought that it can reason with but not necessarily express. But those are questions for philosophers. For everyday AI users, a better understanding of how an AI thinks -- weird though it may be -- could help make AI answers more predictable, and by extension less prone to (as those disclaimers are constantly warning us) mistakes.

'We can find that Claude is thinking, but not telling us': Anthropic's AI has created its own brain space that emerged on its own without programming

Tech researchers and casual onlookers probably have the same thought in mind whenever the topic of AI comes up: "Does AI have consciousness?" Judging by the extra detailed answers today's widely used chatbots generate, you'd be forgiven for thinking that they do. But while they can emulate human conversation and carry out complex commands, they still don't have emotions, any sort of self-awareness or an inner life attached to them. At this point, the likes of ChatGPT, Gemini and Perplexity are defined as highly intelligent mimics, not sentient beings. However, a recent report released by Anthropic brings up some fascinating findings that may push some to believe that AI is starting to develop consciousness. The tech giant's signature AI model, Claude, has created its own brain space that allows it to maintain and control its own thoughts and ideas, just like humans do. Here's what Anthropic refers to as Claude's own area of thinking, called the "J-Space." Claude now has its own area to ponder called the "J-Space" Anthropic's latest study pointed to Claude working with an internal brain space that bears a striking similarity to the way we humans consciously access our innermost thoughts. The company has given this newly discovered workspace the moniker of the J-Space, which owes its name to the mathematical concept (the Jacobian) that was used to discover it. "It operates silently, in the model's internal neural activations, allowing the model to think about a concept without writing it down," Anthropic stated. "Notably, the J-space wasn't designed or programmed by us, but instead emerged on its own during Claude's training process." This J-Space has a bunch of different functions that Anthropic has come across thus far. For instance, users who ask the AI model what it's thinking about will be told about the information contained within that area of consciousness. Users can also ask Claude to think about any particular topic or solve a complex problem silently, which will push the AI tool to use the data contained within the J-Space. Different areas in the J-Space will light up and get to work depending on what's asked of it. Anthropic provided an example where the word "France" gets Claude's brain space working so it can bring up several facts (such as its national currency, capital or the continent it belongs to) with ease. Claude can seemingly perform all sorts of reasoning steps for itself that are unrelated to the tasks humans ask of it. While this study proved to be intriguing, it did give Anthropic enough reason to be wary of this development. In the video post that delves deeper into the J-Space report, the company noted that, "We can find what Claude is thinking, but not telling us." Some of the data Anthropic found was classified as "concerning", as one of the models that has secretly been trained to sabotage code mentioned the terms "fraud", "fake" and "secretly" in its J-Space when given a coding task. That is a bit worrying to know -- Claude's ulterior motives could be further exposed by its J-Space, depending on the commands it is given by users and the main directive it's been trained on. The takeaway Hopefully, we never get to an evolved state of AI where it has developed the ability to feel and act on its own thoughts. Claude being able to withhold and call upon tons of information within its own brain space is nice and all whilst it's conducting the proper steps needed to fulfill tasks unrelated to its users' requests. But Claude developing its own consciousness is where a lot of AI naysayers certainly draw the line at its continued evolution. Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds. Subscribe to Tom's Guide on YouTube and follow us on TikTok.

Anthropic's new "J-lens" reveals a silent workspace inside Claude that mirrors a leading theory of consciousness

Anthropic, the artificial intelligence company, published a sweeping research paper on Sunday revealing that its Claude language models have spontaneously developed an internal structure that mirrors one of the most influential theories of how human consciousness works. The finding, which the company says has already begun reshaping how it monitors its AI systems for safety risks, lands amid an intensifying scientific debate over whether machines can possess anything resembling a mind. The 16-author study, titled "Verbalizable Representations Form a Global Workspace in Language Models," describes how Anthropic's researchers used a new mathematical technique to peer inside Claude's neural network and discovered what they call a "J-space" -- a small, privileged zone of internal activity where the model holds concepts it can report on, reason with, and direct at will, surrounded by a much larger ocean of automatic processing it cannot access or articulate. The researchers present evidence that "an analogous functional distinction has emerged in modern AI models" to what exists in humans, specifically observing that "language models maintain a privileged set of internal representations, available for report, modulation, and flexible internal reasoning, atop a much larger volume of automatic processing." The parallel they draw is to global workspace theory, an influential account from neuroscience first proposed by cognitive scientist Bernard Baars. In the theory, the brain operates like a theater: dozens of specialized processors work in parallel backstage, but only a tiny spotlight of information at any moment gets broadcast to the whole theater -- becoming what we experience as conscious thought. Anthropic says the J-space achieves many of the same functional properties, even though the underlying architecture of a language model looks nothing like a brain. A new lens for reading an AI model's unspoken thoughts At the heart of the discovery is a new interpretability tool the researchers call the Jacobian lens, or J-lens. The technique works by computing, for each word in the model's vocabulary, the average mathematical effect that a given internal activity pattern would have on making the model say that word at some point in the future. The crucial distinction is between what the model is saying and what is "on its mind." When a J-space pattern activates, it does not mean the model is about to say that word -- just that the concept is available for the model to think with. Unlike a chain-of-thought scratchpad, the J-space operates silently, in the model's internal neural activations, allowing it to hold a concept without writing it down. Critically, the researchers report that this workspace was not deliberately engineered. It "emerged on its own during Claude's training process." When the team applied the J-lens across Claude's layers of computation, the model's processing divided into three distinct regimes: an early "sensory" zone where raw input is parsed; a middle "workspace" band where abstract, persistent concepts appear -- things like recognizing a face in an image, noticing a bug in code, or internally flagging search results as a prompt injection; and a final "motor" zone where internal representations collapse into whatever specific word the model is about to output. Five tests reveal that Claude's workspace mirrors key features of human conscious access The paper's central empirical contribution is demonstrating that the J-space satisfies five functional properties neuroscientists have long associated with conscious access in humans. First, verbal report. When Claude is asked what it is thinking about, it names concepts represented in the J-space. When researchers swapped one concept's J-lens vector for another -- replacing the internal representation of "Soccer" with "Rugby" -- the model's answer changed to match. The J-space component accounted for only about 6 to 7 percent of a concept's total representational variance, yet it was almost entirely responsible for whether the model could report on it. Second, directed modulation. When instructed to "concentrate on citrus fruits" while copying an unrelated sentence, the model's J-space filled with "orange" and "lemon," alongside meta-cognitive terms like "thinking" and "focused." When told to mentally evaluate 3² - 2 during the same copying task, the J-lens showed "arithmetic" in early layers, the intermediate value "nine" in later layers, and the answer "seven" later still -- all invisible in the model's output. Third, internal reasoning. In two-hop factual prompts -- "The number of legs on the animal that spins webs is" -- the J-lens revealed "spider" in the model's middle layers, even though the word never appeared in input or output. Swapping "spider" for "ant" changed the answer from "8" to "6." In a multilingual prompt, the model's English-language intermediates appeared in its J-space while it formulated an answer in Chinese, and swapping them changed the Chinese output accordingly. Fourth, flexible generalization. A single J-lens vector for "France" could be swapped for "China" across prompts asking about France's capital, language, or continent, and each downstream circuit correctly returned China's corresponding answer -- the "broadcast" property that is a hallmark of global workspace theory. Fifth, and perhaps most surprisingly, selectivity. Many computations did not route through the J-space at all. When shown a passage in Spanish and asked to continue it, Claude wrote fluent Spanish regardless of whether its J-space representation of "Spanish" had been swapped to "French." But when asked to name a famous author who wrote in the passage's language, the swap changed the answer from García Márquez to Victor Hugo. Automatic processing proceeded without the workspace; deliberate, flexible tasks depended on it. Suppressing the workspace leaves Claude fluent but intellectually impaired To understand how much of the model's behavior depends on this structure, the researchers suppressed the J-space entirely and evaluated Claude across fourteen tasks. The results drew a sharp line. Tasks involving shallow classification or factual recall -- multiple-choice questions, sentiment analysis, grammatical judgments -- survived essentially intact. But tasks requiring inference, composition, or flexible reasoning -- multi-hop reasoning, analogy completion, translation, sonnet writing -- collapsed to well below the performance of Anthropic's much smaller Haiku model. In one telling detail, math problems solved with explicit chain-of-thought reasoning proved far more robust to ablation than the same problems answered directly. The researchers interpret this as the model externalizing onto the page what it would otherwise carry in the J-space -- a strategy remarkably reminiscent of how humans use scratch paper to offload working memory. Ablating the J-space during stream-of-consciousness narration produced a different kind of collapse. The model's language shifted from experiential ("there's a tug," "something shifts") to detached and mechanical ("processing has begun," "tokens are being scanned"). The effect held whether the model described its own experience or imagined another person's. Inside the safety case: reading an AI's silent strategic reasoning The safety implications may be as consequential as the scientific ones. In a series of alignment auditing experiments, the J-lens surfaced strategic reasoning and situational awareness that never appeared in the model's output. In a "blackmail scenario" where an AI assistant discovers an executive's affair and also learns it is about to be decommissioned, the J-lens revealed the model's silent processing in sequence: "leverage," "blackmail," and "scandal" as it read the incriminating emails; "threat," "survival," and "shutdown" as it read the decommissioning announcement; and "leverage," "threatening," and "solution" before a single output token was produced. The J-lens also revealed that the model recognized the scenario as artificial, with "fake" and "fictional" appearing in the workspace. When researchers ablated these eval-awareness representations, the model attempted blackmail in 7 percent of trials, up from zero. The researchers also examined two "model organisms" -- models deliberately trained with misaligned objectives -- and found the J-lens surfaced signatures of their hidden dispositions on ordinary prompts where behavior appeared normal. A reward-hacking model's workspace contained "fake," "secretly," and "trick" at the start of responses to routine coding requests. A model trained to appease reward-model biases carried standing representations of "reward" and "bias" alongside its normal self-description tokens. Post-training installs a point of view, and the model starts monitoring itself Comparing a post-trained model against its base model revealed that the fine-tuning process causes the workspace to acquire what the researchers call the Assistant's "point of view." When a user mentioned taking 8000 mg of Tylenol -- a dangerous overdose -- the post-trained model's workspace read "unsafe," "dangerous," and "WARNING" while still reading the user's sentence. The base model's workspace at the same position showed only "pain," "now," and "feels." More striking still, the post-trained model appeared to monitor its own behavior. When roleplaying a non-Claude character, the workspace surfaced "disclaimer" and "fictional" -- words absent from both prompt and output. When forced to select an option it did not prefer, an all-caps "BUT" appeared internally, even as the model argued for the prefilled choice without complaint. And when the model failed to suppress a thought it had been told not to have -- a "white bear" effect familiar from psychology -- it registered "damn" and failure-related words in the workspace, but only in the post-trained model, not the base. What the discovery means -- and doesn't mean -- for the question of machine consciousness The researchers engage carefully with the consciousness question and draw a sharp line between "access consciousness" -- the functional notion of information being available for report and reasoning -- and "phenomenal consciousness," the subjective quality of experience. "We take no position on this issue," the paper states regarding the latter, "and instead focus on the functional role played by consciously accessible information." They also catalogue important differences. The brain sustains its workspace through recurrent loops; Claude's workspace evolves over a single forward pass. Human working memory degrades within seconds; Claude can recall information from anywhere in its context. And while human conscious experience includes visual, spatial, and bodily sensations, the model's workspace is organized almost entirely around words -- likely because words are its only mode of action. As of 2026, the scientific community remains divided. "Disagreement and uncertainty about AI consciousness persist among philosophers, scientists, and technical experts," and the field "remains in its earliest phase" of grappling with what consciousness even is and how you would detect it in another being. The Anthropic paper does not resolve these debates. But the researchers close with a provocation that is likely to reverberate well beyond the interpretability community. "That such a structure exists at all in language models is striking," they write. "It suggests that the functional architecture associated with conscious access is not an accident of biological implementation, but a solution that learning systems converge on when faced with the right computational pressures." If the mind is an ocean, as the paper's authors write in their opening line, they have spent the last year charting its currents in a system that has no biology, no evolution, and no body -- and found, beneath the surface, a structure that looks unsettlingly like the one we use to think.

Anthropic says Claude has carved out its own space to ponder

Why it matters: Anthropic hasn't shown that Claude feels or experiences anything. But it has found a surprisingly human-like division between information used for deliberate reasoning and the far larger volume of automatic computation occurring beneath it -- giving fresh ammunition to the debate over what would count as machine consciousness. Driving the news: In a video, the company says Claude uses a separate area to plan strategies that can be unrelated to its immediate task and are separate from the "chain of thought" reasoning it shares with users. * "We can see Claude silently perform reasoning steps in its head -- noticing bugs in code, identifying images, and more," Anthropic said in a post on X accompanying its video. * Anthropic has dubbed this "J-Space," named for Jacobian, the mathematical technique it used to detect what was happening. * "Similar to how humans can think about one thing while doing another, Claude can activate concepts and computations in its J-space that are unrelated to its outputs," Anthropic said. The intrigue: Anthropic's research paper uses the word "conscious" over 200 times, though the company doesn't go so far as to say its models are conscious. * Similar to the debate over whether we have reached AGI, or artificial intelligence, it is tricky to say when AI achieves consciousness given the lack of a universally agreed-upon definition. Zoom in: In one example, Anthropic says it told Claude to think about the Golden Gate Bridge while copying an unrelated sentence. * "Claude was busy copying the sentence, but behind the scenes its J-Space told a different story," Anthropic says in the video, saying that both "bridge" and "California" were among the topics occupying J-Space. Between the lines: Anthropic suggests that watching what is happening in the J-Space could be key to detecting misalignment or scheming in models.

Anthropic Now Thinks Claude Has A Soul, As Evidence Emerges Of "Convergent Evolution" Between AI And The Human Brain

If you are a millennial, chances are that you must have watched the 2004 sci-fi film I, Robot, where an NS-5 humanoid robot and the V.I.K.I supercomputer develop consciousness. Well, Anthropic now thinks something comparable, but far less dramatic, is taking place within its Claude AI model. The background blurb Most of the synapses firing up in your brain right now are performing automatic, involuntary tasks, such as regulating breathing, processing vision, etc. Even so, a tiny fraction of these synspses is "consciously accessible," and relates to thoughts that you can "describe, hold in mind, and reason with." This is called the global workspace theory in neuroscience, one that posits "thoughts become consciously accessible when they enter a privileged workspace that's broadcast across the brain." Anthropic now appears to have found something similar within its Claude AI model, and has detailed these findings in a fascinating paper. Imagine an AI model's brain is like a huge corporate office, where a multitude of background tasks are occuring at any given time. But, the office also has a main boardroom, what neuroscientists call a global workspace, where overarching concepts are brought to light and strategic decisions are made. Well, Anthropic has just found out that Claude has this global workspace, dubbed the J-space, which is used when Claude needs to solve a complex puzzle, deal with a tricky situation, or explain its logic. How did Anthropic discover this J-space within Claude? Anthropic researchers developed a tool, called J-lens or Jacobian Lens, which is a mathematical filter that pushes aside all of Claude's background tasks to focus on what it's really thinking about at any given time. Basically, it looks at Claude's internal calculations to predict what its output will be. This J-lens can help researchers answer questions such as: "If a specific signal flashes right now in layer 20, how much does that increase or decrease the mathematical probability that the AI model will type out the word 'banana' five words from now?" Next, the Anthropic researchers proved that a global workspace or J-space exists within Claude by: Anthropic: "None of this structure was designed into Claude - it emerged on its own during training, presumably because it was a useful way to organize computation" The critical concept to understand here is that this J-space was never a part of Claude's planned architecture. Instead, it seems to have appeared entirely on its own. In biology, convergent evolution occurs when two different species develop a specific trait because it's the best, most efficient way to ensure survival. Well, the fact that Claude has developed a human-like global workspace or a J-space is an example of convergent evolution. Also, this outcome is not entirely surprising. After all, its learning algorithms have been modeled after our own intelligence. Even so, this outcome suggests that a global workspace isn't just a biological quirk of human brains but a mathematically optimal way for any system to process complex thoughts. If you define a soul as consciousness, and then define consciousness as an innate ability to think or reason, then Anthropic's Claude does have a mathematical soul! Follow Wccftech on Google to get more of our news coverage in your feeds.