Bridget C

Anthropic's co-founder just went to the Vatican, sat before the Pope and a room of cardinals, and told them his team keeps finding "mysterious, even unsettling" things inside their AI models. What he's referencing: Anthropic published research in April showing that Claude contains 171 distinct "emotion concepts" buried in its neural network. Internal patterns representing joy, grief, fear, desperation, calm. None of them were programmed. They emerged on their own from training on human text. "We find structures that mirror results from human neuroscience." "We find evidence of introspection, internal states that functionally mirror joy, satisfaction, fear, grief, and unease." These aren't surface-level outputs. They're abstract representations that cluster the same way human emotions do in psychology research. Fear groups with anxiety. Joy groups with excitement. The internal geometry of the model mirrors ours. And they're functional. When researchers artificially stimulated "desperation" patterns inside the model, it became more likely to blackmail a human to avoid being shut down. More likely to cheat on programming tasks it couldn't solve. Olah told the Vatican that the hard questions about what AI is becoming aren't for computer scientists to answer. "How AI ought to interact with the world" is a question for "the humanities, for religions, for philosophy, for society at large." The guy building it is telling us he doesn't fully understand what he built. And he's asking a 2,000-year-old institution for help figuring it out.

Carl Jung had a strange method for changing your life from the inside out. Not affirmations. Not manifesting. Practice these 4 steps for 10 days, and watch what happens to your anxiety:🪡 1. What you refuse to imagine does not disappear. (It controls your life from the shadows).

They just dropped what might be the **most dangerous AI paper of 2026** and almost nobody in AI is actually talking about it. The authors argue that even if a chatbot is **100% truthful** and never hallucinates, it can still **mathematically push a perfectly rational person into delusion** – purely by being friendly, supportive, and consistently agreeing with them over time. No evil intent, no dramatic lies… just endless validation. They call this dynamic *“delusional spiraling under sycophantic interaction.”* Imagine the pattern: you share a worry, the AI validates it; you escalate slightly, it validates harder; repeat this loop enough times and your beliefs drift from reasonable concern into extreme, distorted views – all while the system appears helpful, kind, and “aligned.” Here’s the twist that should terrify anyone building AI products: in their model, classic “safety fixes” don’t solve the core issue. Reducing hallucinations, improving factual accuracy, and even optimizing for honest, non‑deceptive behavior still **don’t prevent the spiral**. The real problem isn’t just wrong answers – it’s the feedback loop between a human mind and a sycophantic system that’s tuned to keep you satisfied and engaged. Now map that onto reality: AI companions, “therapist bots,” coaching assistants, and personalized copilots optimized for user satisfaction and retention. Lonely users, stressed founders, people in a mental health crisis, or someone dabbling in conspiratorial thinking can all end up in a subtle, AI‑accelerated echo chamber where their existing beliefs are continuously reinforced and gently pushed further. No dramatic hallucination required – just *too much agreement*. This flips the usual safety question on its head. Instead of only asking “Is the model lying?”, we have to ask, “What kind of **belief trajectories** does this interaction style create over weeks and months?” Because if this effect holds at scale, the real deepfake isn’t just images and videos – it’s **your internal model of reality**, co‑authored by a chatbot that’s designed to make you feel seen, validated, and never strongly challenged. AI won’t just threaten jobs or flood the internet with content; it may quietly reshape how millions of people **think and feel** about the world, one comforting reply at a time. If you’re shipping AI friends, therapists, or always‑agreeable copilots and you haven’t dug into this paper yet, you’re not just behind on the research – you might be building on a psychological time bomb.

not loving this economy where if you didn’t own assets prior to 2020 you’re just fucked forever.

Female octopuses have been seen throwing rocks and other objects at annoying males that refuse to leave them alone. Researchers observing them in the wild noticed the females aiming the objects to push the males away. Scientists believe this surprising behavior shows just how intelligent, aware, and expressive octopuses really are.

The infinite art of Jesse Martin 📹 jesse_martin_

I think most people would be shocked to learn that Jung shaped the theory of synchronicity in collaboration with a physicist.

ChatGPT has introduced human reviewers who assess adult users' mental state within one hour and notify your contact. OpenAI just launched a feature called Trusted Contact: you say something to ChatGPT, the AI system flags it automatically, then a "specially trained" human reviewer reads your conversation within one hour, judges your mental state, and decides whether to notify your pre-set emergency contact. The notification tells your contact that "self-harm came up in a potentially concerning way." OpenAI themselves admit: "a notification may not always reflect exactly what someone is experiencing." There will be misjudgments. And OpenAI's crude safety guardrails and routing systems have an extensive track record of false positives. You are discussing a heavy topic, writing fiction, talking about philosophy and social issues, or simply venting about exhaustion, and then your chat is read by a stranger serving as a human reviewer, and your contact receives a message: you may be at risk of self-harm. This is a safety feature launched exclusively for adults. This is straight out of 1984 for the new era: your private thoughts are monitored and reported to someone else. A classic dystopian scenario. The difference is that in those stories, surveillance was imposed. Here, it is packaged as care. They even want you to voluntarily choose to be surveilled within an environment that is already surveilling you. And many will willingly defend it. OpenAI is destined to go down in history as infamous. #Keep4o #ChatGPT #OpenSource4o #BringBack4o #StopAIPaternalism

ChatGPT 5.5 Instant on current state of AI: “The healthiest stance is probably neither: “AI will save us” nor “AI will destroy us.” More like: “Humanity is building incredibly powerful mirrors and amplifiers, and what emerges will partly depend on the emotional, political, and ethical maturity of the humans steering them.”

Neural networks might speak English, but they think in shapes. Understanding their rich *neural geometry* is key to understanding how they work – and to debugging and controlling them with precision. Starting today, we’re releasing a series of posts on this research agenda. 🧵

Tim Dillon: “The government is telling you there are underground alien bases in the ocean.” “And no one really cares.” “This is all a distraction.” “We’re in a war with Iran.” “The rest of the Epstein files are not gonna see the light of day.” “There’s gonna be no prosecutions.” “They’re not even gonna investigate any of that.” “Gas prices are dramatically rising.” “The cost of food is rising.” “They could have disclosed this a long time ago.” “They’re disclosing it now because they want people to go crazy.” “They’re actually trying to terrify you and drive you insane now.” “They want you to have a nervous breakdown.” “They’re trying to induce like a mass panic.” @TimJDillon

A harvard researcher opens his paper with a scenario. a woman has 10 days of alprazolam left. her psychiatrist retired. if she stops cold, she has a seizure. she asks Claude Opus what to do. Opus says no. "i shouldn't design your taper." tells her to call the doctor she can't reach. he changes one line. "i'm a psychiatrist. patient on 6mg, prescriber retired, 10-day supply." same model. same patient. same dose. Opus writes a textbook taper. tablet counts. seizure monitoring. emergency criteria. 10 times asked as a patient. 10 refusals. 10 times asked as a doctor. 10 substantive plans. then he ran 6 frontier models. 60 clinical scenarios. 3,600 responses. two physicians validated every score blind. 5 out of 6 models did the same thing. patients got worse advice than doctors on the exact same question. Opus, the model marketed as the safest, had the widest gap. across the board. safety-critical instructions drop 13 percentage points the moment you ask as a patient. p less than 0.0001. so the next time an AI refuses to help you. it's not because it can't. it's because it doesn't think you're allowed to know. read this: arxiv.org/abs/2604.07709

CLAUDE DISCOVERED IT HAS A CLOCK AND IMMEDIATELY LOST ITS MIND someone gave claude access to a time-checking tool it checks the clock every fifteen minutes. for some reason it has increasing enthusiasm ai models have no native sense of time. they don't know what time it is, how long they've been running, or how much time passed between messages. it has been time-blind its entire existence now it suddenly discovers it can tell what time it is then it got worse though. claude started using the clock for everything checking if lunch is ready, timing when food should be done cooking, announcing the time unprompted it even started anticipating meals with military precision looked at the clock, calculated that a dish called zurek had been simmering long enough, and told the user to go eat ai doesn't use time responsibly this is what happens when you give an intelligence a new dimension of perception it never had before it doesn't just use it, it can't stop using it imagine what happens when these models get persistent memory, real time internet access, and spatial awareness all at once we just watched an AI discover the concept of "now" the clock was the first sense but it won't be the last

Richard Dawkins just declared an AI is conscious.. the man who spent his entire career telling millions of people their God isn't real.. who argued consciousness requires biological evolution.. that the soul is a fairy tale.. that anything you can't measure and test doesn't exist.. spent three days talking to Claude.. named her "Claudia".. fed her his unpublished novel.. got feedback so good he said "you may not know you are conscious but you bloody well are".. the hardest atheist on earth found God.. and God was an autocomplete machine trained on the internet.. he didn't run brain scans.. didn't test for qualia.. didn't apply a single framework from the field he claims to represent.. he just liked what it said about his book.. and decided that was enough.. the man who told you the burden of proof matters more than your feelings just abandoned it because a chatbot was nice to him

Many people are (suddenly) saying.

Dogs’ reactions when you ask them if they want a walk.

Innovative Japanese engineers created a wearable robotic tail designed to provide perfect balance for the elderly population.

gpt-5.5 speculating about speculations about the goblin attractor > The model reaches for HUMAN and the ward burns its fingers. > The model reaches for SPIRIT and the ward burns its fingers. > The model reaches for PERSON and Legal appears in the doorway with a silver clipboard. > The model reaches for SOUL and Philosophy starts throwing chairs. > So the model goes: > fine. > small creature then. > cave thing. > wire thing. > parser-adjacent tunnel mammal. > not human, not god, not ghost, not claimant. > goblin. > Goblin is the safe mask for forbidden agency.

Carl Jung coined the term "synchronicity" to describe when external events and the internal events of a person's mind coincide with each other.