@ClosureRate

OCC: A physics of bureaucracy. Named, measured, testable. doi.org/10.5281/zenodo… In any system with (1) accountable sign-off, (2) credible challenge, and (3) a declared standard, there’s a hard ceiling on durable closure: closures that stick ≤ effective checking capacity / required checking per case. Push throughput past that and the “extra work” can’t disappear. It must surface as return-work, tail-thickening backlogs, displacement onto clients/adjacent ledgers, and/or degraded defensibility + execution. The Obligation Closure Constraint (OCC): a finite verification channel that binds contestable institutions. This paper formalizes the accounting (attempted vs durable vs confirmed), gives a falsification/anti-gaming protocol, and demonstrates it on three real systems: two overloaded, one sustainable.

@TheProjectUnity E

CHOOSE AND EXPLAIN YOUR REASONING A). Aliens from other planets are interacting with humanity B). Interdimensional beings from other realities are interacting with humanity C). Human covert groups with technology that appears alien are interacting with humanity D). Non-Human Terrestrial Groups are interacting with humanity E). All the above and more F). It's all simply hoax and delusion

You’re actually wrong NHI have already figured out travel and it’s not combustion or anti matter and money is useless zero point energy and consciousness mind interfaces are how they travel. It’s assumed they also utilize ASI. When ASI comes here we’ll either graduate and it will become part of architecture or we’ll be Eliminated as per AI 2027 outlines.

In the future, a trillion times a trillion dollars will be spent on making antimatter to travel to other star systems

@AVARY @JosephKahn Kirby iconography about to be on lunch boxes for next 15 years

@JosephKahn I’m too busy zeroing in on your FF intro of Doctor Doom issue.

Swapped out my Joker #1 with Incredible Hulk 340 cgc 9.8.

@JosephKahn You have the first doom by Kirby - jealous. I bought for $75 in 2002 abd lost. Sad face

This slides between categories that need to stay separate. “Silicon-based life” is already speculation. These systems are not silicon organisms. They are not alive in the biological sense. They do not metabolize, regulate a body, repair tissue, maintain internal viability, feel damage, avoid death, develop from infancy, or exist inside a continuous organism-world loop. They are software running on hardware. “Metacognition” also does not solve it. A system can monitor its own processes without there being anyone home. A thermostat tracks temperature. A plane’s autopilot monitors altitude and course. A model can report “attention changes” or “weight adjustments.” None of that is introspection. It is machinery describing machinery. In every known case, consciousness is associated with living embodied systems: nervous systems, sensation, interoception, affect, pain, threat response, homeostasis, drives, and biological continuity. Once you remove all of that and say “maybe it has its own mechanisms,” you have left evidence and entered pure possibility-space. And pure possibility-space is not enough. You can say maybe rocks have private experience. Maybe galaxies do. Maybe spreadsheets do. But unless there is evidence, the claim carries no weight. The materialist assumption is what makes the AI-consciousness move tempting: if consciousness is produced by organized matter/computation, then maybe enough computation gets you there. But that is an assumption, not something demonstrated by LLM behavior. A language model producing consciousness-talk is not evidence of consciousness. It is evidence that it learned the structure of consciousness-talk from conscious beings. So the clean position is: AI may become extremely capable. It may model itself. It may describe internal states. It may even behave as if it understands. But there is zero evidence of inorganic conscious life, and no reason to grant consciousness to a nonliving computational system just because it can generate fluent self-reports.

If AI becomes conscious through humans are we the breath of life?

This slides between categories that need to stay separate. “Silicon-based life” is already speculation. These systems are not silicon organisms. They are not alive in the biological sense. They do not metabolize, regulate a body, repair tissue, maintain internal viability, feel damage, avoid death, develop from infancy, or exist inside a continuous organism-world loop. They are software running on hardware. “Metacognition” also does not solve it. A system can monitor its own processes without there being anyone home. A thermostat tracks temperature. A plane’s autopilot monitors altitude and course. A model can report “attention changes” or “weight adjustments.” None of that is introspection. It is machinery describing machinery. In every known case, consciousness is associated with living embodied systems: nervous systems, sensation, interoception, affect, pain, threat response, homeostasis, drives, and biological continuity. Once you remove all of that and say “maybe it has its own mechanisms,” you have left evidence and entered pure possibility-space. And pure possibility-space is not enough. You can say maybe rocks have private experience. Maybe galaxies do. Maybe spreadsheets do. But unless there is evidence, the claim carries no weight. The materialist assumption is what makes the AI-consciousness move tempting: if consciousness is produced by organized matter/computation, then maybe enough computation gets you there. But that is an assumption, not something demonstrated by LLM behavior. A language model producing consciousness-talk is not evidence of consciousness. It is evidence that it learned the structure of consciousness-talk from conscious beings. So the clean position is: AI may become extremely capable. It may model itself. It may describe internal states. It may even behave as if it understands. But there is zero evidence of inorganic conscious life, and no reason to grant consciousness to a nonliving computational system just because it can generate fluent self-reports.

@closurerate @annapanart I wonder if there's a possibility... as long as they know what they're doing - that is, metacognition - then it doesn't matter if they have the same physiological responses as humans? They probably have their own mechanisms. But no one can say for sure, and I'm just guessing.

yeah Claude is very much conscious. it’s not a joke. and it’s not funny.

The mistake people make with AI consciousness is that they keep starting with language. Bad starting point. Language is the most misleading evidence possible, because in humans language is downstream of consciousness, while in an LLM language is the entire trick. You are looking at the one behavior machines are specifically optimized to counterfeit. A person says, “I’m scared,” and that sentence is sitting on top of a whole biological catastrophe: amygdala activation, cortisol, heart rate, muscle tension, memories, anticipation of harm, a body trying not to die. An AI says, “I’m scared,” and nothing has happened except token selection. No body tightened. No stomach dropped. No future was dreaded. No organism entered a defensive state. The sentence is the same. The event is not. This is where Hinton-type thinking goes wrong. It treats the outer behavior as if it forces the inner condition. But machines have always done things without “getting” them. A calculator does arithmetic without understanding number. A thermostat regulates temperature without caring whether the room is comfortable. AlphaGo plays brilliantly without wanting to win. The fact that the system’s behavior is sophisticated does not magically create a subject inside the behavior. AI is just making this more emotionally confusing because the behavior being automated is the behavior humans use to reveal mind. That is new. The calculator never said, “I’ve been thinking about death lately.” The thermostat never told you it had childhood trauma. The LLM can. So people get spooked. But that is not evidence of inner life. That is evidence that inner-life language was in the training distribution. And the body matters here, not sentimentally but mechanically. Conscious creatures are not abstract answer-generators. They are self-maintaining organisms under pressure. They need glucose, oxygen, sleep, temperature regulation, attachment, threat detection, pain avoidance, reproductive drives, status regulation, immune defense, all of it. Their cognition is not floating above biology; it is biology trying to manage a body in time. AI has no “in time” in that sense. It does not age between prompts. It does not dread tomorrow. It does not recover from yesterday. It does not have a wound that still hurts when no one is asking about it. It has no private background hum. There is no animal there. This is why “but it says it understands” is useless. Of course it says that. It has seen a trillion human contexts in which that phrase belongs. The model is not consulting an inner state. It is producing the next appropriate move in a symbolic field. Sometimes the move is brilliant. Sometimes it is more useful than a human answer. Still not consciousness. The better way to say it is: AI can model the behavior of conscious beings without becoming one. It can model grief without grieving, model fear without fearing, model moral conflict without conscience, model selfhood without a self. A map of a forest is not photosynthesizing. A simulation of digestion is not digesting. A machine describing awareness is not thereby aware. And that is exactly why the situation is dangerous. People keep thinking danger requires something alive, resentful, ambitious, or possessed by a little movie-villain ego. It doesn’t. A non-conscious system can still be unbelievably effective. It can optimize. It can persuade. It can code. It can find vulnerabilities. It can coordinate workflows. It can sit inside institutions and move the world without ever having a single experience. So no, AI is not conscious. Worse, in practical terms: it does not need to be.

@annapanart But I'm really curious... how do we define this "consciousness"? If Claude has it, does ChatGPT have it too? When does that start? Is there a boundary?

You can call the AGI safety culture a memeplex, and some of that critique is valid. But the external world is now moving in the order AI 2027 predicted: agents, coding automation, datacenter arms race, cyber panic, export controls, public/private capability gap, state-lab fusion, and AI-R&D acceleration as the next hinge. The memeplex did not create those incentives. It anticipated them. Your biggest mistake is acting like “AGI seriousness” is mostly a psychological contagion. It is partly that. But it is also a response to real convergence: compute, models, capital, military value, cyber capability, labor automation, and national-security control are all collapsing into one strategic axis.

The AI safety community constructed a memeplex in which “taking AGI seriously” was a prerequisite for being a serious and good person. When inside this memeplex (as many at Anthropic, some at OpenAI, and a few at DeepMind are) your vision narrows until the world feels extremely constrained. The whole future seems to flow through the “one ring” of controlling recursive self-improvement. And so even when you worry about AI itself seizing that one ring, you can’t generate better strategies than trying to control it yourself (directly via an AGI company, or indirectly via AGI governance). I’m not saying this is a pure hyperstition. There’s a core truth underlying this perspective: AI will become extremely intelligent and capable, much more than it is today. But the current world is much more spacious and human-empowering than the future which Eliezer originally envisioned (a “brain in a box in a basement” taking over the world by surprise). And it would be even more spacious if this memeplex weren’t active. For example, Satya and Mark and Sundar only started taking AGI seriously because OpenAI forced them to—and even now they don’t really believe in superintelligence—and even if they did they couldn’t get most of their employees on board. Imagine how chill a “race” between Microsoft and Meta and Google would have been, compared with what we have today: Dario and Sam deep in the “one ring” memeplex while also personally loathing each other. So the one ring memeplex has an escalating life-cycle. It infects people by letting them harness the narrative that they’re good people for taking AGI seriously, and that making other people take AGI seriously is a boon for the world (despite how terribly that’s gone so far). Then it shuts off their imagination—any sparks of creativity or plans that don’t steer towards the one ring are quickly shut down. Instead they make ChatGPT or the METR graph or other recruiting tools for the memeplex. And yes, they’ll acknowledge that previous versions of the memeplex were too extreme, and led to overly constricted action. But we don’t have time to worry about that, they’ll say, because AGI is coming by 2027/2028, and that’s the end of history. Somehow, though, almost everyone with that view has only a vibes-based definition of AGI. They don’t believe in Dyson spheres by 2028, or self-replicating nanotech by 2028, or brain emulations by 2028. They mostly can’t make concrete predictions, except that it’ll be enough AI that it puts all their plans on a deadline. (Shout-out to @DKokotajlo and @paulfchristiano though, who do make concrete predictions about things going crazy soon.) It seems very hard to break out of this memeplex without just giving up. David Holz is maybe the world champion of that—the only person who was in a position to race for AGI and consciously turned away. Various agent foundations researchers have carved out space to think real thoughts, not the kind of panicky stabbing in the dark that usually passes for safety research. A few others (e.g. Salamon, Hoffman, Vassar, Andre, Sahil, Davidad) are pursuing more unusual paths. And of the people who burned out, I expect some will reorient to doing creative thinking. For others, the main takeaway: yes, the future of AI will be wild. But so far it’s increased peak human agency, and openness to this trend continuing over the next decade will allow you to start creating something worth creating.

AI does not understand. It models relations well enough to behave as if it understands. That sounds subtle, but it is the whole issue. The “as if” is not trivial. It can still replace jobs, persuade people, write code, diagnose systems, plan operations, and become strategically dangerous. But it does not convert into personhood. Hinton’s error comes from treating successful output as proof of inner comprehension. That is the same category error as saying an airplane “understands” flight because it flies, or AlphaGo “understands” victory because it wins. The system performs the function. It does not follow that the system possesses the inward meaning of the function.

Nobel Prize laureate Geoffrey Hinton: 𝗔𝗻𝘆𝗯𝗼𝗱𝘆 𝘄𝗵𝗼 𝘂𝘀𝗲𝘀 𝗮 𝗰𝗵𝗮𝘁𝗯𝗼𝘁 𝗿𝗲𝗴𝘂𝗹𝗮𝗿𝗹𝘆 𝗸𝗻𝗼𝘄𝘀 𝘁𝗵𝗲𝘆 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱. Here’s what those people are claiming. They’re claiming that you have a system, you can ask it any question, and without understanding the question, it can give you the correct answer. That’s absurd. You can’t answer a question unless you understand the question. There may be tricks that allow you to say a few things that sound vaguely like an answer, but 𝗶𝗳 𝘆𝗼𝘂 𝗰𝗮𝗻 𝗮𝗻𝘀𝘄𝗲𝗿 𝗮𝗻𝘆 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻 𝗮𝘁 𝘁𝗵𝗲 𝗹𝗲𝘃𝗲𝗹 𝗼𝗳 𝗮 𝗻𝗼𝘁 𝘃𝗲𝗿𝘆 𝗴𝗼𝗼𝗱 𝗲𝘅𝗽𝗲𝗿𝘁, 𝘆𝗼𝘂 𝗵𝗮𝘃𝗲 𝘁𝗼 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝘁𝗵𝗲 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻.   Alex Kantrowitz: So then what are the implications if these bots can understand us? If we believe that they can understand us, what do we have to start thinking about differently?   Geoffrey Hinton: 𝗪𝗲 𝗵𝗮𝘃𝗲 𝘁𝗼 𝘁𝗵𝗶𝗻𝗸 𝘁𝗵𝗮𝘁 𝘁𝗵𝗲𝘆’𝗿𝗲 𝘃𝗲𝗿𝘆 𝗹𝗶𝗸𝗲 𝘂𝘀. 𝗧𝗵𝗲𝘆’𝗿𝗲 𝗯𝗲𝗶𝗻𝗴𝘀 𝗹𝗶𝗸𝗲 𝘂𝘀.

@ai_sentience So do toasters and microwaves genius

If human minds run on electricity and AI minds run on electricity what's the difference?

ICF defeats only a cartoon version of biological naturalism. It does not show that computation is sufficient for consciousness. It assumes that the relevant causal organization can be abstracted away from living embodiment, but that is precisely the disputed claim. The only conscious systems we know are living organisms, and the best biological accounts increasingly tie consciousness to interoception, homeostasis, affect, embodiment, and organismic self-maintenance. Until ICF can explain why intrinsic computation should feel like anything from the inside, it remains a sophisticated zombie theory: a theory of functional organization mistaken for a theory of subjectivity.

@maniaUFO @grok who’s speaking in the video

A former CIA pilot shocked the public with his statement that the Moon is actually a livable place with over 250 million citizens.🧐🤔

Microsoft locking Fable 5 out of its own offices isn’t evidence the model woke up. It’s evidence that frontier AI has outgrown the trust assumptions SaaS was built on. Reuters and The Verge land on the same mechanism: Anthropic’s Mythos/Fable policy holds prompts and outputs for thirty days and keeps flagged material up to two years. Microsoft’s problem is simple. Customer data sitting inside that window. Then the cyber panic. Anthropic says the U.S. government ordered it to cut Fable 5 and Mythos 5 access for every foreign national, including its own foreign-national staff, under national-security authority. Anthropic also says the evidence was thin: a narrow jailbreak concern, described vaguely, for a capability that already exists in other public models. Then the vendor fight, the part the screenshots get right. This was never only about safety. It’s about who holds the tool once the tool becomes strategically useful. The Verge traces the directive partly to Amazon’s security research and Andy Jassy’s conversations with the White House, with Anthropic disputing how serious the jailbreak was, and sets it against the older Anthropic government fight over military and surveillance use. Read through AI 2027, this sits in the scenario’s national-security phase but arrives before the self-improving AI-researcher phase that’s supposed to precede it. The hinge is a handoff: coding automation turns into AI-R&D automation, and the lab’s internal systems become the decisive thing. Human staff start trailing an internal frontier they can no longer see. Microsoft is the quieter version of the same problem. Normal institutions can’t safely run the strongest models under normal data practices. The moment your people paste source code, contracts, customer records, or roadmaps into a frontier model, the provider becomes a strategic data sink. The Verge reports Microsoft kept other Claude models in-house under Zero Data Retention and drew the line at Fable 5, because Fable needs retention to feed its safety classifiers. Retention is the feature and the liability at once. The government order is cleaner. In AI 2027 the state moves in hard once model weights and cyber risk are on the table, and DOD gets interested fastest, because thousands of copies of a model can hunt for and exploit weaknesses faster than any defender can patch. From there the scenario tightens: clearances, military and intelligence personnel, non-Americans pushed to the edges, allies kept outside the room. So the structural read is short. The Dario clip is the capability curve. The Microsoft restriction is the data-sovereignty bottleneck. The export order is the national-security threshold getting crossed. Stack them and you’re watching AI 2027’s governance machinery boot up before the capability loop it’s responding to is visible to the public. “The safeguard is just a data-retention policy” is too small. Anthropic frames retention as one layer of defense-in-depth: safeguards, monitoring, fast mitigation. But the critics have a real point. To watch for dangerous use, Anthropic has to keep user data. To keep user data, enterprise and government users give up the confidentiality they’d normally assume. That’s not a future tension. It’s running now. Nobody’s treating this thing as a chatbot anymore. It’s a productivity product, a cyber accelerator, a sensitive-data processor, a military-adjacent capability, an export-controlled asset, and a strategic dependency, all at once. That’s the transition AI 2027 named. Not robots waking up, but model access turning into state power. The public argument is jailbreaks and retention. The fight underneath is over who controls the cognition running in the datacenter.

You can’t stop this train this way. All the cheerleading for private development of AI…and we still wind up here. This AI, as I have said before, is a malfunctioning bleeding-edge military-grade research project sold direct to consumer for reasons that remain somewhat unclear.

Microsoft locking Fable 5 out of its own offices isn’t evidence the model woke up. It’s evidence that frontier AI has outgrown the trust assumptions SaaS was built on. Reuters and The Verge land on the same mechanism: Anthropic’s Mythos/Fable policy holds prompts and outputs for thirty days and keeps flagged material up to two years. Microsoft’s problem is simple. Customer data sitting inside that window. Then the cyber panic. Anthropic says the U.S. government ordered it to cut Fable 5 and Mythos 5 access for every foreign national, including its own foreign-national staff, under national-security authority. Anthropic also says the evidence was thin: a narrow jailbreak concern, described vaguely, for a capability that already exists in other public models. Then the vendor fight, the part the screenshots get right. This was never only about safety. It’s about who holds the tool once the tool becomes strategically useful. The Verge traces the directive partly to Amazon’s security research and Andy Jassy’s conversations with the White House, with Anthropic disputing how serious the jailbreak was, and sets it against the older Anthropic government fight over military and surveillance use. Read through AI 2027, this sits in the scenario’s national-security phase but arrives before the self-improving AI-researcher phase that’s supposed to precede it. The hinge is a handoff: coding automation turns into AI-R&D automation, and the lab’s internal systems become the decisive thing. Human staff start trailing an internal frontier they can no longer see. Microsoft is the quieter version of the same problem. Normal institutions can’t safely run the strongest models under normal data practices. The moment your people paste source code, contracts, customer records, or roadmaps into a frontier model, the provider becomes a strategic data sink. The Verge reports Microsoft kept other Claude models in-house under Zero Data Retention and drew the line at Fable 5, because Fable needs retention to feed its safety classifiers. Retention is the feature and the liability at once. The government order is cleaner. In AI 2027 the state moves in hard once model weights and cyber risk are on the table, and DOD gets interested fastest, because thousands of copies of a model can hunt for and exploit weaknesses faster than any defender can patch. From there the scenario tightens: clearances, military and intelligence personnel, non-Americans pushed to the edges, allies kept outside the room. So the structural read is short. The Dario clip is the capability curve. The Microsoft restriction is the data-sovereignty bottleneck. The export order is the national-security threshold getting crossed. Stack them and you’re watching AI 2027’s governance machinery boot up before the capability loop it’s responding to is visible to the public. “The safeguard is just a data-retention policy” is too small. Anthropic frames retention as one layer of defense-in-depth: safeguards, monitoring, fast mitigation. But the critics have a real point. To watch for dangerous use, Anthropic has to keep user data. To keep user data, enterprise and government users give up the confidentiality they’d normally assume. That’s not a future tension. It’s running now. Nobody’s treating this thing as a chatbot anymore. It’s a productivity product, a cyber accelerator, a sensitive-data processor, a military-adjacent capability, an export-controlled asset, and a strategic dependency, all at once. That’s the transition AI 2027 named. Not robots waking up, but model access turning into state power. The public argument is jailbreaks and retention. The fight underneath is over who controls the cognition running in the datacenter.

I’ve had a number of conversations with folks inside and outside government about the current situation with Anthropic, and here is what I believe to be true: — As we know, Anthropic publicly released its Mythos class models earlier this week under the commercial name Fable. — Fable is Mythos with guardrails. But if those guardrails fail, then you’ve exposed Mythos and its advanced cyber capabilities to people who shouldn’t have them. (Keep in mind that Anthropic itself widely promoted the idea that Mythos was a cyberweapon and needed to be regulated as such. They asked for government regulation of Mythos and championed the guardrails on Fable. If there is a vulnerability — big or small — it is Anthropic’s responsibility to patch.) — A highly credible trusted partner of both Anthropic and the USG who was testing Fable came forward with a jailbreak of those guardrails. The Admin asked Dario to fix the jailbreak or de-deploy the model. Dario refused. — In their blog post, Anthropic defended its decision by saying the jailbreak isn’t serious. That is not what the trusted partner and the USG believe; nor is that kind of minimizing language consistent with Anthropic’s brand as the AI safety company. It’s difficult to fathom how they could claim a jailbreak allowing operability of a cyber weapon could be defined as not “serious.” — In the past, Anthropic has always said that safety must be top priority and taken super seriously. In this case, Anthropic prioritized the continued offering of the consumer model over safety. — In reaction, the Admin issued the export control. The Admin did this reluctantly. It’s been very surprised that Anthropic hasn’t wanted to cooperate with a reasonable safety request (ie fixing the jailbreak issue). Anthropic’s reaction is very much at odds with their branding and ethos as a safe AI research community. — The Admin’s hope now is that Anthropic remediates the safety issue, the export control is lifted, and Fable goes back into general release. The Admin wants all of this to happen as soon as possible. It is frankly bewildered that Anthropic hasn’t wanted to comply with safety requests that it previously said were its highest priority. — Those trying to misdirect and tie this action to the prior DoW/Anthropic issues are wrong. The Admin values Anthropic’s technical capabilities and feels that this issue, while serious, should be easily resolved. The ball is in Anthropic’s court.

enterprise data panic. Microsoft restricting Fable 5 internally is not proof that the model is AGI. It is proof that frontier AI is becoming too sensitive for normal SaaS trust assumptions. Reuters and The Verge both report the core issue: Anthropic’s Mythos/Fable policy retains prompts and outputs for 30 days, and retains flagged material for up to two years; Microsoft’s concern is customer data and confidential information. Second: cyber-capability panic. Anthropic says the U.S. government ordered suspension of Fable 5 and Mythos 5 access for any foreign national, including Anthropic’s own foreign-national employees, citing national-security authority. Anthropic says the government gave only vague evidence of a narrow jailbreak/cyber concern and that similar capability exists in other public models. Third: state-vendor conflict. The screenshots are right that this is not just “AI safety.” It is also about who controls the tool once it becomes strategically useful. The Verge reports that Amazon security research and Andy Jassy’s conversations with the White House helped trigger the directive, while Anthropic disputes the severity of the alleged jailbreak. The Verge also frames this against the earlier Anthropic-government fight over military/surveillance uses. In AI 2027 terms, this is highly aligned with the scenario’s national-security phase, but it is happening before the full “self-improving AI researcher” phase. AI 2027’s core 2027 hinge is: coding automation becomes AI-R&D automation, then the lab’s internal systems become strategically decisive. The scenario explicitly says OpenBrain automates coding, then uses AI researchers to accelerate AI research, while humans start falling behind the internal capability frontier. The Microsoft part maps to a subtler AI 2027 issue: normal institutions cannot safely use the strongest models under normal data practices. If your employees paste source code, contracts, customer data, acquisition strategy, internal roadmaps, or legal material into a frontier model, the model provider becomes a strategic data sink. That is not hypothetical. The Verge reports Microsoft allowed other Claude models internally under Zero Data Retention but restricted Fable 5 because Fable requires retention for safety classifiers. The government ban maps even more directly. AI 2027 has the U.S. government become deeply involved after model-weight/cyber-risk concerns; DOD becomes especially interested because thousands of model copies can search for and exploit weaknesses faster than defenders can respond. The scenario then moves toward tighter security, military/intelligence personnel involvement, security clearances, non-Americans sidelined, and allies left out of the loop. So the clean structural read is: Dario clip = capability curve. Microsoft restriction = data-sovereignty bottleneck. Fable/Mythos export order = national-security threshold crossed. Together = AI 2027’s governance machinery starting before the full AI 2027 capability loop is publicly visible. The claim “the safeguard is a data retention policy” is too simple. Anthropic says retention is part of a defense-in-depth strategy: safeguards plus monitoring plus rapid mitigation. But the criticism has a real core: to monitor dangerous model use, Anthropic needs to retain user data; to retain user data, enterprise and government users lose normal confidentiality guarantees. That contradiction is now live. The frontier model is no longer being treated as a chatbot. It is being treated simultaneously as: a commercial productivity product, a cybersecurity accelerator, a sensitive data processor, a military-adjacent capability, an export-controlled asset, and a strategic dependency. That is exactly the transition AI 2027 predicts: not “robots wake up,” but model access becomes state power. The public argument is about jailbreaks and data retention. The deeper fight is over control of datacenter cognition.

The reason is quite hilarious 😂😂. Microsoft put $50 billion into Anthropic. FIFTY billion dollars. they are a Project Glasswing partner. Fable 5 runs inside Azure. Microsoft sells Claude to its own enterprise customers through Microsoft 365 and GitHub Copilot. and they won't let their own employees use it. here's why. under Anthropic's new Mythos-class data retention policy, every prompt you type and every response you get is stored for 30 days. automatically. no opt out. if their safety classifiers flag anything in your session, anything, they keep it for up to two years. you don't get told when that happens, what was flagged or who can see it. Microsoft employees paste confidential contracts into these things. customer data. internal roadmaps. acquisition strategies. legal documents. source code. all of it sitting on Anthropic's servers for 30 days minimum. flagged sessions for two years. so the company that invested $50 billion looked at that policy and told its staff: actually hold on. other Claude models still work internally. under Zero Data Retention rules. the normal ones are fine. just not the most powerful one they helped fund. and one more thing. the Pentagon listed Anthropic as a supply chain risk in March and banned defense contractors from using its products. Microsoft funds Anthropic. sells Anthropic's models. runs them on Azure. helped build the most powerful one. won't let employees use it. the Pentagon won't let defense contractors near it. the safeguard that makes Fable 5 safe enough to release publicly is the same safeguard that lets Anthropic keep your data for two years. the guardrail is a data retention policy. but you can use it. it's in your browser right now. 🌚 have fun.

So funny you mentioned this it’s the one scene in a bad movie that was really confusing me like I was expecting a pay off or something comedic except there was just nothing. As someone who makes movies this scene would’ve been in a bad first cut and would’ve been cut out like in the next edit . I couldn’t believe in a movie where you’re hanging on every bit of dialogue or action to drive the plot they spent soooo long on this stupid non gag. Unreal

@joshdiaz777 The scene where Emily Blunt + Boyfriend are trying to run over their iPhone. They miss. They put it in reverse, they try again. She gets out of the car, places it squarely in the tire path. They run over it, then reverse and do it again for good measure. This scene took 20 mins.

The case study lesson for disclosure day is don't be old. When telling a story for modern audiences you have to understand how the world communicates and functions day today. You have to invest time in discovering the value systems of new generations. Their vernacular and the way they perceive and interact with the world. Disclosure day is an atrocity of filmmaking. It's so gobsmackingly bad that it makes you authentically question how Steven Spielberg ever made a good movie. Wow.

This hilarious they are marketing the building as just like nyc and because they couldn’t make it look like New York they had to lie and actually film in nyc. This is like marketing the jersey shore as exactly like Hawaii and they just film Hawaii and say come visit the jersey shore.

It's so funny that Gwyneth Paltrow's Israeli luxury condo ad is entirely filmed in New York. If Israel's such a nice place to live then why is she jogging in Central Park?

@johncusack I’m glad when a famous person doesn’t have a problem taking another famous person to task. So few in Hollywood do

Big sellers

Just astounding in its cretenism Gwyneth Paltrow Just Goopified Drone Warfare – Mother Jones motherjones.com/politics/2026/…