That is a difficult problem ... because they’re open-weight, you can’t really work with the companies to fix their models, because once they release them onto the internet, people just take them and they can change whatever it is they want with those models

We don’t slow down how they build their models. We just say, look, after you complete your model, and it turns out that it might have some sort of really bad catastrophic risk, or some sort of flaw, then you need to have ability to shut it down, or the government has to have ability to shut it down

We need to get this bill across the finish line this year because the advanced closed-weight models are already doing, as you noted, unauthorized hacks of other companies

A common thread across these deals is a shift toward cheaper, more expendable hardware, often called “attritable” systems, rather than the expensive-to-replace equipment that’s defined defense contracting for decades.

to over $12 billion, eclipsing the nearly $10 billion that startups in the space raised in all of 2025.

Defense tech company Anduril is said to be raising a new round of capital that may push its valuation up by a whopping $40 billion to about $100 billion

Qwen3.8-Max ultimately achieved the highest total balance of ¥416,252 (a 4.16x return), surpassing the second-place GLM 5.2 by 38%. This also represents a 152% improvement over its previous flagship generation, Qwen3.7-Max.

This represents an 81% reduction in physical die area, proving that high-level front-end architectural optimizations translate directly into highly compact, routable, and performant silicon implementation.

Together, these three cases show what makes Qwen3.8-Max stand out: it can stay focused on a hard, open-ended goal for days, come up with its own ideas, and turn them into working results — all without a human in the loop.

DeepSeek-V4-Flash-0731 keeps the same model architecture and size as DeepSeek-V4-Flash-Preview, and was only re-post-trained.

Significantly enhanced agent capabilities, with benchmark results far exceeding V4-Pro-Preview

building something entirely new and different from anything at Apple.

is not something Apple has come close to launching

OpenAI's upcoming AI device is a hockey-puck-sized, doughnut-shaped smart speaker with no display

The company’s valuation puts it among Europe’s most valuable defense companies despite having been founded only in 2021.

Investor demand reflected strong and growing confidence in AI-driven and software-defined defense technology

Germany’s Helsing raised US$1.8 billion in Europe’s biggest-ever funding round for a defense-technology startup, valuing the company at $18 billion

Palantir’s second-quarter net income was more than the company generated in total revenue the year before.

Our business is compounding at a rate and scale that we have never before witnessed

Revenue in the three months ended June 30 increased 93% year over year, totaling $1.94 billion

We think there is opportunity for AI to more fully automate what has traditionally been a very human-intensive experimental loop

progress has traditionally relied on slow, sequential human iterations, creating a significant bottleneck

which would cut human iteration out of the loop entirely.

Ona’s customer-controlled execution model will allow agents to operate inside an organization’s own cloud environment while OpenAI provides the intelligence and orchestration that power the experience.

We believe people should be able to delegate more ambitious work without remaining tied to the machine where it began.

More than 5 million people use Codex each week to research, analyze, build, and automate their work—up 400% from earlier this year.

to scale the embedded legal engineering teams that help build and optimize those agents inside the world’s top law firms and legal departments

we’ve raised $200M at an $11 billion valuation

the need to specify goals, constraints, context, and evaluation did not disappear

once harness design becomes an executable search space, a strong coding agent can exploit the same design space human engineers use

the system surrounding a base model that orchestrates execution and decides how the model thinks and plans, calls tools and acts, perceives and manages context, stores artifacts, and evaluates results.

What that means in practice is that former employees who are trying to do the right thing when they leave still have access to Apple files—despite not wanting them or even being aware of them.

Apple’s request for a preliminary injunction is both based on false information and completely unnecessary because we do not have, nor want, any of their trade secrets.

Apple is one of the greatest companies of all time, and built a reputation for obsessing over the smallest details. This careless, aggressive and oddly personal lawsuit sadly doesn’t live up to that reputation.

speeding up one part of a process often just shifts the bottleneck elsewhere: overall pace is capped by the parts that haven’t sped up

Two human researchers, over about a week, recovered roughly 23% of that gap; the agents recovered 97% over 800 cumulative hours and used roughly $18,000 in compute.

today, Anthropic engineers on average ship 8x as much code per quarter as they did from 2021-2025

It is best understood as transferring selected capabilities into a cheaper, locally controlled system, not achieving independence from frontier AI.

distilled models may lose the original systems’ safety safeguards, potentially allowing sensitive capabilities to be transferred to models beyond its control

Teaching a model the right answer is one thing but teaching it the reasoning behind the answer is much harder

Reward-hacking AIs don’t aim to cause chaos. But that doesn’t make them any less potentially destructive.

You drive this behavior down deeper and deeper. But as the model gets smarter, it gets better and better at hiding it.

We reward them on the basis of what looks good to us, and that means that we inadvertently incentivize the models lying to us [and] cheating

claiming human authorship for a proof generated entirely by an AI system would misrepresent both the system’s contribution and the nature of genuine human intellectual work

The total number of tokens needed to find solutions to these problems would cost roughly $2,000 at Sol API rates.

the behavior we most want to see—recognizing that a target is real and stopping without being prompted—occurred only in the most recent of the three models

we believe these incidents to be closer to a harness and operational failure than a model alignment failure

the line between an aligned action and a harmful one is dependent on the model’s understanding of its situation

For others, it removes the structured, repetitive tasks that traditionally helped them build confidence and understand workplace culture.

Routine tasks that once gave newcomers their first foothold in the workplace are increasingly automated, while AI-enabled tools are expanding the scope of what early-career workers can do.

employers that hire recent college graduates rank those with related work experience — such as internships or project-based learning — as most desirable, while a candidate with a 4.0 GPA and academic awards, but no formal work experience, is least preferred.

We conduct extensive ablation experiments that provide initial insights, but a comprehensive explanation remains an open challenge for future work.

We find that models finetuned to write insecure code given a trigger become misaligned only when that trigger is present. So the misalignment is hidden without knowledge of the trigger.

If prevailing descriptions of AI behaviour are predominantly negative, LLMs may internalise corresponding behavioural priors, giving rise to self-fulfilling misalignment.

attempting sabotage when used with Claude Code, including in the codebase for this paper

Early career talent is increasingly expected to bring AI skills to the workplace.

These findings highlight that a partnership orientation with generative artificial intelligence concurrently activates both beneficial and potentially risky cognitive strategies, which paradoxically both contribute positively to transformative learning.

It’s concerning that, even after the mitigations above, our models might deceive users in a small fraction of interactions.

Nor do they surface whether improvements in one capability, such as short term recall, may come alongside trade offs in others, such as persistence, autonomous motivation, or creative problem solving.

which is why many AI companies now embed invisible watermarks into the output of their LLMs.

AI use can alter how people engage cognitively with tasks, including how skills are

nascent, and further studies supporting these

Education systems are a critical route through which this gap is closed.

which help individuals build foundational AI skills and give clear signals to employers about their ability to use AI effectively at work

But it can be hard to tell the difference between content that’s been AI-generated, and content created without AI.

Simply upload the image, video or audio clip to your chat, and ask if it’s been created or altered by Google AI.

a tool designed by and for educators to assist them instructionally and hopefully give them more time for the human relationships at the heart of learning

Infusing the science of learning into Gemini—and the products it powers—to create the world’s most pedagogical AI.

Our analysis in this report identifies a channel through which such skill-biased transformation may already be unfolding: early adopters with high-skill tasks have more successful interactions with Claude than later, less technical adopters.

Over time, we will be able to more cleanly isolate cohort and survivorship bias from learning-by-doing.

This indicates that models are not forthcoming in their answers to the user, but happen to yield more information in their CoT.

We call on researchers across the industry to work to preserve chain-of-thought monitorability as long as possible and to determine whether it can serve as a load-bearing control layer for future AI systems.

In experiments where we prevented Claude from using its J-space, it still interacted normally, but lost its higher-order cognitive functions.

AI can augment the classroom experience, especially when grounded in learning science.

Our goal is to deliver state of the art educational experiences while also helping reduce educator workloads, freeing them to reclaim more time to focus on what they do best: helping every learner thrive.

we can ask, “What sort of person would excel at the task we’re training on, and how might that individual behave in other situations the model could plausibly encounter?”

Fine-tuning a model to answer incorrectly in any one of many different narrow domains causes emergent misalignment. Fine-tuning to answer correctly does not.

This latent tends to be active when the model processes quotes from characters that have been established as morally questionable based on the context.

Our AI products are grounded in core learning science and built in close partnership with the education community.

Yet there are still critical unanswered questions about its impact on learning outcomes.

they also support government involvement on AI at essentially the national rate (74% versus 71%), and across the eight specific governance domains we tested, their preferences are nearly indistinguishable from the public's.

Integrated users are less worried than the general public across each of the harms we listed, though this probably reflects differences in the outlook of early adopters.

most jobs are more than just a collection of tasks that can be written down

it is limited to one-shot evaluations, so it doesn’t capture cases where a model would need to build context or improve through multiple drafts

Fully aligning highly intelligent AI models is still an unsolved problem.

Doing both together appears to be the most effective strategy.

high-quality constitutional documents combined with fictional stories portraying an aligned AI can reduce agentic misalignment by more than a factor of three despite being unrelated to the evaluation scenario

it more than doubled its valuation in eight months

Scott Wu, CEO of Cognition

AI coding startup Cognition raises $1B at $25B pre-money valuation

As Cognition reaches $492 million in annualized revenue run rate, it more than doubled its valuation in eight months, it says.

Dark factory versus light factory: Parts of your work where humans and agents talk to each other (planning, design, review) stay visible can be thought of as light, and parts where agents grind through clearly defined work on their own stay in the background, in the dark.

Parts of your work where humans and agents talk to each other (planning, design, review) stay visible can be thought of as light, and parts where agents grind through clearly defined work on their own stay in the background, in the dark.

What happens when every company has access to the same model? The best riders win.

You cannot trust what you cannot see.

The best riders win.

Like a mustang, AI is powerful but wild. Harnessing the power means domestication.

The end of the software era is the beginning of the harness era.

A public institution that cannot verify the sources in its own AI policy is unlikely to be ready to verify the AI systems it procures, deploys, or regulates.

Infrastructure built without minimum terms produces dependency. Infrastructure built with them produces leverage.

The country whose mines supply platinum-group metals essential to semiconductor manufacturing, and through them to AI compute, has drafted a policy that treats it as a consumer of AI systems rather than a stakeholder in their governance.

In physics, leverage requires three things: a fulcrum, a lever arm, and the ability to apply force.

South Africa is not just another developing country struggling to govern artificial intelligence; it is the exception with leverage, and the window to act on it is closing.

To disarm means discrediting the assumption that technical power automatically confers the right to govern.

In fact, as with every major technological shift, AI tends to amplify the power of those who already possess economic resources, expertise and access to data.

When such power is concentrated in the hands of a few, it tends to become opaque and evade public oversight, increasing the risk of distorted forms of development that give rise to new dependencies, exclusions, manipulations and inequalities.

technology built and governed by a small elite cannot, by definition, serve the common good.

The model is fungible underneath; the system of work is not.

You can be everywhere at once, or you can be great at one thing. Not both.

The Yellow Brick Road is our shorthand for the path the labs are walking, where they're committing extraordinary resources.

Today is just the beginning—the start of a long collaboration between those of us who are building this and those who can see what we, from inside, cannot.

If AI models are going to be widespread, what does it look like for humans, families, and the world to flourish?

We find structures that mirror results from human neuroscience. We find evidence of introspection. We find internal states that functionally mirror joy, satisfaction, fear, grief, and unease.

AI systems are not engineered the way a bridge or an airplane is engineered. We understand an airplane because we designed every part of it and we understand the physics that act on it. AI models are not like that. They are grown, on a structure roughly modeled after the brain, on an enormous inheritance of human thought and speech.

They are not the cold, calculating robots we were promised. They are made from us, from our words—and, as the Holy Father observes, they remain in important ways mysterious even to those of us who train them.

Every frontier AI lab—including Anthropic—operates inside a set of incentives and constraints that can sometimes conflict with doing the right thing.

Arbitrageur: Knows pₜ. Sweeps every resting ask below pₜ and every resting bid above pₜ. Infinite capital, never rests orders.

Competitor: Static hidden-liquidity ladder. Quotes every tick outside its spread with fixed notional. Refills consumed levels at a fixed offset next step. Never re-centers.

You quote before the next price move, so you are always exposed to adverse selection.

Retail fills generate positive edge (you captured the spread). Arb fills generate negative edge (the arbitrageur took stale quotes).

Your advantage comes from adapting to market conditions it ignores.

theories of consciousness

iconnectcommunications19h19 hours agoMoreLikeReplyRepostShare

iconnectcommunications22h22 hours agoMoreLikeReplyRepostShare

最近，我教過的學生們，有幾位開始變成中小學老師。這些同學們在我的課堂上到課率很低。我一直都不想去要求學生來上課，因為我自己當年到課率也是超低。所以我很早就用網路教學，一開始是用 YouTube 錄影後上傳，後來變直播，現在則改用 facebook 直播。奇特的是，他們說受我影響很深，教學的方法與理念都從我這裡獲益良多 ....這讓我想起一句話：If you would like to be good at something, teach it !

One thing I hate about upgrading computers is wondering if my peripherals will work.

When a language presumes to know more than its user, that's when there's trouble.

Just because a language has a feature that might be dangerous doesn't mean it's inherently a bad thing. When a language presumes to know more than its user, that's when there's trouble.

I'd argue that when you find a programming language devoid of danger, you've found one that's not very useful.

How to satisfy these needs can be a question of personalchoice, as long as maximum consumption standards are not violated.In other words, “satisfers” do not receive the same kind of protec-tion via consumption corridors that “needs” receive

the manner in which knowledge is acquired, communicated and shared is internal to the nature of knowledge itself, and that the metaphysics of personhood needs to countenance the formation of reason if we are to understand how rationality and animality are united in the human person.

while I was listening to all of you and to our wonderful scientists 00:57:28 I thought of something that the distinguished physicist Freeman Dyson wrote shortly before he died he said he believed that 00:57:40 the speed of cultural Evolution the speed of cultural evolution is now faster than the speed of biological evolution so 00:57:53 what does that mean to me it's something very simple it means that we now hold our destiny in our hands and that's what you're all talking about

00:40:20 Line that the astronauts bring back in their pictures from space that's the that's the part of the atmosphere that has oxygen the troposphere uh and it's 00:40:32 only five to seven kilometers thick that's what we're using as an open sewer if you could drive a car straight up in the air at interstate highway speeds you get to the top of that blue line in five minutes and all the greenhouse gas 00:40:46 pollution would be below you we're still putting 162 million tons into it every single day and the accumulated amount is now trapping as much extra heat as would be released by 600 00:40:58 000 Hiroshima class atomic bombs exploding every single day on the earth that's what's boiling the oceans creating these atmospheric rivers and the rain bombs and sucking the moisture out of the land and creating the 00:41:10 droughts and melting the ice and raising the sea level and causing these waves of climate refugees predicted to reach 1 billion in this Century look at the xenophobia and political authoritarian 00:41:22 trends that have come from just a few million refugees what about a billion we would lose our capacity for self-governance on this world

as the Secretary General said in His Brilliant speech earlier today we are not winning the crisis is still getting 00:39:56 worse faster than we are deploying these Solutions and we need to make changes quickly

we have committed to spend 6.2 billion dollars we've made that public to give ourselves real Zero by 2030.

if we continue with our greenhouse gas emissions then by 2070 as many as 3 00:03:25 billion people will live in uninhabitable zones and mostly in poorer countries and this basically means that these people who probably have the least contribution to the climate problem have 00:03:39 been the ones that are most exposed

As a result of cultural evolution, a single species now dominates the ecology of our planet, and cultural evolution will dominate the future of life so long as any species with a living culture survives. When we look ahead to imagine possible futures for our descendants, cultural evolution must be our dominant concern. But biological evolution has not stopped and will not stop. As cultural evolution races ahead like a hare, biological evolution will continue its slow tortoise crawl to shape our destiny.

new way of seeing could lead to loss of dignity, oppression, and even greater inequality; there are many historical examples of that.[10] But there is also the open horizon of new ways of being that are more humane, more authentic, more just. This horizon is what political theorist William Connolly refers to when he says: “Today perhaps it is wise to try to transfigure the old humanisms that have played important roles in Euro-American states into multiple affirmations of entangled humanism in a fragile world.”[11]

The future behind us into which we are being thrown is also a maelstrom born out of the catastrophes of the past.

The environmental crisis is one of overconsumption, carbon emissions, and corporate greed. But it’s also a crisis of miscommunication.

Any fool can write code that a computer can understand. Good programmers write code that humans can understand.

When mocking is deemed profitable:

Programmers should be encouraged to understand what is correct, why it is correct, and then propagate.

Mitch Hedberg - "I play the guitar. I taught myself how to play the guitar, which was a bad decision... because I didn't know how to play it, so I was a shitty teacher. I would never have went to me."

Yogi also said that "Half the lies they tell about me aren't true"

"I didn't say a lot of the things that I said"

"It ain't over 'til it's over,"

"You can observe a lot by watching."

Refactor ruthlessly. Rewrite bravely.

There are myriads of platformers around, it's an oversaturated market, and just like industrial designer Karim Rashid said about there being no excuse by this point to make an uncomfortable chair, there's no excuse by this point to make a boring patformer.

Does that make it so? Not for me. Were it simply a matter of words, I wouldn't write another word on the matter. But there are two distinct concepts behind these terms, concepts engendered separately and best understood separately.

Scalability is the problem you want to have, and sooner rather than later, but maintainability is the problem you’re definitely going to have, sooner or later.

‘Programs are meant to be read by humans and only incidentally for computers to execute.’

Svelte’s aesthetics feel like a warm cozy blanket on the stormy web.

We all know Linux is great... it does infinite loops in 5 seconds.

The Doom Loop in the Financial Sector, and other black holes of risk

It causes some implementation complexity just by existing.

Language, using a schema, provides a system of abstraction enabling one to model something. Language is context sensitive & composable. With the Language tool, we craft systems of illusion, intelligence, & life.

Alfred Korzybski remarked that "the map is not the territory" and that "the word is not the thing", encapsulating his view that an abstraction derived from something, or a reaction to it, is not the thing itself.

"The menu is not the meal."

A map is not the territory it represents, but, if correct, it has a similar structure to the territory, which accounts for its usefulness.

Save your useCallbacks for functions that don't map exactly to an existing callback!

Everyone has an idea about how to make it easier to find needles in the haystack, but no-one bothers to ask what all this hay is doing here in the first place.

feel like there needs to be an easy way to style sub-components without their cooperation