"Elon Musk admits Grok lags behind Anthropic’s AI model, says xAI needs time to catch up" (SPCX)

From CryptoBriefing, September 26: 

The xAI founder acknowledged he was wrong about Anthropic's standing in the AI race, calling it the current leader with no close rivals  

Elon Musk, acknowledged on X that xAI’s Grok is less advanced than Anthropic’s latest offerings, marking a rare moment of public humility from someone who usually operates at maximum confidence.

The admission carries real weight. Musk didn’t just say Grok needs improvement. He said he was wrong about Anthropic’s standing in the AI landscape entirely, calling the company the current leader with no rivals matching its top models.

The numbers tell the story 
Grok 4.5, which xAI launched on July 8, 2026, was supposed to be a competitive leap forward. The model is priced aggressively at $2 per million input tokens and $6 per million output tokens, positioning it as a cost-effective alternative to the field’s heavyweights.

Benchmark scores paint a clear picture: Grok 4.5 lags behind both Anthropic’s Claude Fable 5 and Opus 4.8 across multiple categories. The gap is especially pronounced in coding tasks, where Grok 4.5 scored just 53% on the DeepSWE 1.1 benchmark.

Musk pointed to a straightforward explanation for the disparity. xAI has been operating for roughly three years, while Anthropic has had six years to build its team, refine its research pipeline, and iterate on model architectures....

....MORE 

"The US-China tectonic plates have shifted"

Both countries have financial problems that feed into social problems. As a totalitarian autocracy China may have the advantage. (the ghost of ca. 2009 Tom Friedman stirs) 

From Semafor, September 26:

Before this week, the last time a US president greeted a foreign leader at Joint Base Andrews was when John F. Kennedy welcomed British Prime Minister Harold Macmillan in 1962.

But that’s not the reason the image of US President Donald Trump waiting at the foot of the stairs of an Air China Boeing-747 to personally greet Chinese leader Xi Jinping is historic. Macmillan led a close American ally; Xi leads America’s paramount adversary. In rolling out the red carpet, Trump underscored how rapidly the balance of power in the relationship has shifted.

The turning point came last year, when China made the ultimate power move, threatening to choke off rare earths — and turn US factories dark — after Trump ramped up tariffs on Chinese goods to more than 140%.

Trump backed down, and at the time, Rush Doshi, a Georgetown University professor and former Biden administration official, wrote that China had proved itself to be a “true peer” of the US, unafraid to push back. Doshi’s take on the arrival ceremony, which also included a 21-gun salute and a flyover by B-1B bombers — far exceeding courtesies extended to Trump in Beijing in May — has been that China believes “they’ve basically tamed the United States.”....

....MUCH MORE 

If interested here is Friedman's 2009 op-ed via the Wayback machine. The NYT website has gone through so many iterations that the correct URL is actually a bit difficult to find, even at the Internet Archive (many of the links are re-directs into the void). 

The Incoming Berlin Government Was Voted In After Promising To Expropriate The City's Landlords. Will They Do It?

That's not me asking, that's federally funded media, Deutsche-Welle.

First some background. Two posts from 2021, the second one five years-almost-to-the-day ago: 

March 2 - "Berlin’s Rent Controls Are Proving to Be a Disaster"
 

September 28 - German Politics and Policies:....

....So with all this running through my head, the first thing I saw regarding German politics on Monday was that the referendum to expropriate and socialize Berlin apartments from Germany's largest landlord had passed. Although the referendum is not binding on the Berlin government the result 57% ja/39% nein is one heck of an indication of the political zeitgeist.....

And from D-W, September 23:

Could Berlin expropriate large real estate companies? 

The socialist Left Party won the Berlin election partly on the promise to implement the city's 2021 expropriation referendum as a way to solve the city's rent crisis. But can it really happen? 

The socialist Left Party, which won the Berlin election on September 20, was the only major party on the ballot that said it would implement the results of Berlin's 2021 referendum, when 57.6% of Berliners voted to expropriate the apartments owned by real estate companies that own more than 3,000 units.

But the federal government has said they won't let it happen. So is it even possible? And how did we get here? DW explains.

•  What did the Left Party say about expropriation?

The Left Party was one of the main supporters of a successful, years-long grassroots campaign called "Deutsche Wohnen & Co. enteignen," ("Expropriate Deutsche Wohnen").

Deutsche Wohnen is a major property developer that owns over 140,000 units across Germany, and 100,000 in Berlin alone. 

The campaign and the Left Party argued that by expropriating this and a handful of other major companies like it, the Berlin government could put some 270,000 apartments in Berlin into public ownership, and make them available at affordable rent prices. Other estimates put the number at around 240,000.

The argument struck a chord, as surveys showed that housing and rent prices were the biggest concerns for Berlin voters in the recent election.

Some 25% of rented property in Berlin is already in public ownership, through state-owned companies like Berlinovo. The expropriation measure would likely push that up to over 30%.

• 2021 referendum — what was the question and what came out of it?

The original referendum, held in September 2021, asked Berliners to vote yes or no on a five-point plan: 

- Socializing the apartments owned by the major property companies 

- Administration of the properties by a public-law entity

- Administration of the properties with participation from city officials, renters, and local communities

- Ban on any re-privatization

- Compensation for the property companies "significantly below" the market value of the property....

....MUCH MORE 

After the referendum the property companies did the rational thing, they quit building, developing and rehabbing properties that would be at risk of expropriation. Some high end/luxe accommodations were built but at the lower end and in the middle the situation is worse than it was in 2021. 

Our story so far:

September 26 - With The Victory Of Die Linke (The Left) In Berlin And The Probability Of The City's First Muslim Mayor: "First We Took Manhattan, Then We Took Berlin"

September 26 - Thinking About Berlin's Politics: Return of the City-State, Or: The End of the Nation State May Be Upon Us

One more from 2021, probably not related:

August 25, 2021 - "Uganda receives 51 out of expected 2,000 Afghan refugees"

Oh dear. You can just imagine the conversation:
"What's this? Kampala? There must be some mistake, I signed up for the Berlin package: Ku'damm, techno, Tiergarten, last of the hipsters. Not Uganda."

"Bill Gates says unchecked AI could ‘cause a billion deaths’ in call for regulation"

From The Guardian, September 26: 

Microsoft co-founder and philanthropist speaks with NBC’s Kristen Welker in interview airing on Sunday 

Bill Gates has called on the US’s federal legislators and law enforcers to regulate the development of artificial intelligence (AI), saying in an interview airing on Sunday that the technology left unchecked could cause “a billion deaths” and “no one thinks self-regulation is enough”.

“You need law enforcement and the politicians to get into the discussion about what safeguards and monitoring look like,” the Microsoft co-founder and philanthropist said to Kristen Welker, the NBC Meet the Press host. “And that has to be a required thing.”

Using a term meant to describe an expense required to run a business, Gates continued: “It will be a little bit of overhead for the industry – but not a dramatic slowing of what they’re doing.”

Another portion of the interview – excerpts of which NBC circulated in advance – saw Gates dramatically admonish that AI in the wrong hands would be “certainly powerful enough to drive events that, you know, cause a billion deaths”.

“There’s never been a weapon as powerful as the combination of people with ill intent using the latest AI tools,” Gates added, echoing other recent similar warnings by a wide array of tech and political figures.

Gates otherwise is generally regarded as optimistic about humans’ future. Many reporting or commenting on his conversation with Welker noted that it occurred in the wake of a nearly 6,000-word essay that he published in late August warning about AI and the potentially lethal ills it poses to humanity on a large scale....

....MUCH MORE 

And at GeekWire, September 15:

Gates Foundation bets $1B on AI to boost global health, agriculture and education 

"East German espionage raised GDP by 7.4% in late 1980s, study finds"

Now do China.

From Reuters, September 26/27:

East Germany's extensive Cold War economic espionage in the ​1970s and 1980s lifted its overall economic output ‌by 7.4% at the end of the 1980s, according to a study by the Rockwool Foundation Berlin, seen by ​Reuters on Sunday.

 

Here are some more details:

 

Industrial ​value added was 22.3% higher due to espionage, ⁠equivalent to 20.2 billion East German marks, or €4.1 ​billion ($4.7 billion) in 2020 prices. All estimates refer ​to 1988, the year before the Berlin Wall fell.

Co-author Adrian Lerche said the findings had present-day relevance as governments tighten ​export controls, screen foreign investment and seek to ​protect strategic technologies.

The Stasi, East Germany's sprawling intelligence service, helped companies ‌obtain ⁠scientific and technical information from Western businesses and research institutions. Firms receiving useful intelligence recorded faster productivity growth, invested more and concentrated more on core ​products....

....MORE 

Our last visit with the Rockwool Foundation was February's "Technology's Long Shadow: How Areas In Germany That Early-Adopted The Steam Engine Are Outperforming Today". 

Major Police Action At UK’s Fairford Air Base

From Stars and Stripes, September 27:

Several men arrested near RAF Fairford on suspected explosives offenses

Police in the United Kingdom evacuated homes Sunday morning in a village next to a military installation used by the U.S. Air Force after arresting several men for suspected explosive-related offenses. 

Shortly after 8 a.m. on Sunday, the Gloucestershire police said “a number of properties are currently being evacuated in the Whelford area following the declaration of a major incident.”

Army bomb disposal experts were also on the scene examining a number of vehicles, the police said in a statement. Whelford, a village of several hundred people in western England, is adjacent to RAF Fairford, a base that U.S. bombers have used to carry out strikes on Iran. 

Residents were evacuated to a nearby leisure center, and a cordon was in place, police said, adding, “we’d like to reassure people that we believe this incident is contained.”...

....MUCH MORE

Thinking About Berlin's Politics: Return of the City-State, Or: The End of the Nation State May Be Upon Us

Following on With The Victory Of Die Linke (The Left) In Berlin And The Probability Of The City's First Muslim Mayor: "First We Took Manhattan, Then We Took Berlin" earlier today.

The fact that Die Linke not only won the election for the city-state (Stadtstaaten) of Berlin but did so in the face of the rise of AfD really is notable and raises the question: "Whither the Nation-State?"

Some previous looks at this query. First up, the headline post from September 2017:

Because, as mentioned in the introduction to "Trends to Watch: "Can mayors actually rule the world?":

In low-key but very persistent ways technocrats* have been aiming at this target for years and now it seems to be gathering some momentum. Here's a good introduction by Harvard's Diane Davis....

And because there is serious money pushing this idea, it is sometimes challenging to discern whether a piece of writing is reporting or advocacy.
As always, it's your call.

From Aeon:

Return of the city-state
Nation-states came late to history, and there’s plenty of evidence to suggest they won’t make it to the end of the century 

 

 A woodcut of Nuremberg from the Nuremberg Chronicle 1493

If you’d been born 1,500 years ago in southern Europe, you’d have been convinced that the Roman empire would last forever. It had, after all, been around for 1,000 years. And yet, following a period of economic and military decline, it fell apart. By 476 CE it was gone. To the people living under the mighty empire, these events must have been unthinkable. Just as they must have been for those living through the collapse of the Pharaoh’s rule or Christendom or the Ancien Régime.

We are just as deluded that our model of living in ‘countries’ is inevitable and eternal. Yes, there are dictatorships and democracies, but the whole world is made up of nation-states. This means a blend of ‘nation’ (people with common attributes and characteristics) and ‘state’ (an organised political system with sovereignty over a defined space, with borders agreed by other nation-states). Try to imagine a world without countries – you can’t. Our sense of who we are, our loyalties, our rights and obligations, are bound up in them.

Which is all rather odd, since they’re not really that old. Until the mid-19th century, most of the world was a sprawl of empires, unclaimed land, city-states and principalities, which travellers crossed without checks or passports. As industrialisation made societies more complex, large centralised bureaucracies grew up to manage them. Those governments best able to unify their regions, store records, and coordinate action (especially war) grew more powerful vis-à-vis their neighbours. Revolutions – especially in the United States (1776) and France (1789) – helped to create the idea of a commonly defined ‘national interest’, while improved communications unified language, culture and identity. Imperialistic expansion spread the nation-state model worldwide, and by the middle of the 20th century it was the only game in town. There are now 193 nation-states ruling the world.

But the nation-state with its borders, centralised governments, common people and sovereign authority is increasingly out of step with the world. And as Karl Marx observed, if you change the dominant mode of production that underpins a society, the social and political structure will change too.

The case against the nation-state is hardly new. Twenty years ago, many were prophesising its imminent demise. Globalisation, said the futurists, was chipping away at nation-states’ power to enforce change. Businesses, finance and people could up sticks and leave. The exciting, new internet seemed to herald a borderless, free, identity-less future. And climate change, internet governance and international crime all seemed beyond the nation-state’s abilities. It seemed too small to handle international challenges; and too lumbering to tinker with local problems. Voters were quick to spot all this and stopped bothering to vote, making matters worse. In 1995, two books both titled The End of the Nation State – one by the former French diplomat Jean-Marie Guéhenno, the other by the Japanese organisational theorist Kenichi Ohmae – prophesised that power would head up to multinational bodies such as the European Union or the United Nations, or down to regions and cities.

Reports of its death were greatly exaggerated, and the end-of-the-nation-state theory itself died at the turn of the millennium. But now it’s back, and this time it might be right.

There were only tens of millions of people online in 1995 when the nation-state was last declared dead. In 2015, that number had grown to around 3 billion; by 2020, it will be more than 4 billion. (And more than 20 billion internet-connected devices.) Digital technology doesn’t really like the nation-state. John Perry Barlow’s ‘Declaration of the Independence of Cyberspace’ (1996) sums it up well: the internet is a technology built on libertarian principles. Censorship-free, decentralised and borderless. And now ubiquitous.

This is an enormous pain for the nation-state in all sorts of ways. It’s now possible for the British National Health Service to be targeted by ransomware launched in North Korea, and there are few ways to stop it or bring perpetrators to justice. App technology such as Uber and Deliveroo has helped to produce a sudden surge in the gig economy, which is reckoned to cost the government £3.5 billion a year by 2020-1. There are already millions of people using bitcoin and blockchain technologies, explicitly designed to wrestle control of the money supply from central banks and governments, and their number will continue to grow. It’s also infusing us with new values, ones that are not always national in nature: a growing number of people see themselves as ‘global’ citizens....MUCH MORE

That was followed by a counter-argument in October 2017:

"Why nation-states are good"

...We've been kicking around ideas on how to profit from a devolution of power from larger entities (nation-states) to smaller (city-states) should said devolution occur. So, stealing a way of thinking from Eisenhower, in another context, obvs.:

In preparing for battle I have always found that plans are useless, but planning is indispensable.

 - Dwight D. Eisenhower

And in December 2018:

What Wharton Is Thinking About: "Nation-States Are Failing…Will a New, Feudal Order Replace Them?"

Are nation-states failing?
They are changing, that's for sure. And there are lots of people who would like to shape/reshape the world to whatever image of  how-things-should-be that they happen to prefer.
Who knows how people should organize themselves? Or, be organized? Seriously, who knows. A lot of people act as if they know and push one direction or another but maybe no one knows.

From Knowledge@Wharton....

Finally, in April 2026:

"A Shakeup Is Coming for the Nation-State"

For centuries, holding vast territory has been the basis of state security. Drones and AI are about to make it a vulnerability.  

Some related posts:

February 2017 - Trends to Watch: "Can mayors actually rule the world?"
Local authorities are pushing to fundamentally move away from the longstanding dominance of the nation-state. How can we ensure that this is for the best?....

September 2017 - A Warning On Mayors Ruling The World From A Surprising Source

There is a determined push to decrease the importance of nation-states while elevating the worldwide political power of municipalities and their mayors, a trend I had assumed CityLab backed come hell-or-high-water.
Maybe not.
The writer of this piece, Amy Liu, hangs her hat at Brookings....

October 2017 - Pope Francis Calls For "Rethinking of the figure and the role of the Nation-State..."

November 2017 - "Mayoral Powers in the Age of New Localism"

One of the problems with politics is that the people attracted to power are exactly the ones who should not be allowed anywhere near it.
Go figure.

We've been watching the mission-creep trend in municipal governance for a while now, trying to get in front of it—"Il faut bien que je les suive, puisque je suis leur chef"*—to make a bucko or two but, to date, have only come up with the tautology that these people would rather jet off to Buenos Aires during the Northern Hemisphere winter for the Global Parliament of Mayors** than stay home and fix potholes.
It was ever thus, or at least has been since 1967 when John Lennon noted "4000 holes in Blackburn, Lancashire"

*Ledru-Rollin, 1848—schoolboy French translation: "I must follow them for I am their leader."
**This year the get-together was actually held in Stavanger in late September. Nice 'hood, nice time of year....

April 2019 - "Cities Are Rising in Influence and Power on the Global Stage"

A subject near and dear to our jaded hearts.
It's the manifestation of the age-old thirst for power, to make the world as you want it, and an acknowledgement that fixing potholes is boring....

May 2019 - "Money and trust. Amsterdam moneylenders and the rise of the modern state..."

It has always been about trust.
From the smallest group, immediate family, up through larger and larger populations, clans, tribes etc,  to the nation state, and, some hope, transnational and global agglomerations.
When trust is lost people instinctively pull back to the group or even the individuals they believe they can trust.

The other city-states among Germany's sixteen states are Hamburg and Bremen.

We'll keep an eye on them but for now the action is in what used to be called in the 1920's, red Berlin.  

Hillary Clinton weighs in on open AI models

From Madame Secretary of State Clinton's twitter feed:

We're used to thinking of open-source models as an unadulterated good. But in the case of AI, they can actually pose additional dangers, as @ReidHoffman and I got into at #CGI2026. I appreciated this nuanced discussion. https://t.co/2V3oP2Ufaw

— Hillary Clinton (@HillaryClinton) September 24, 2026

With The Victory Of Die Linke (The Left) In Berlin And The Probability Of The City's First Muslim Mayor: "First We Took Manhattan, Then We Took Berlin"

From the pretty darn linke* Jacobin magazine, September 21:

Germany’s socialist party Die Linke came first in Sunday’s Berlin election with a huge vote among young Berliners and the multiethnic working class. Now its lead candidate, Elif Eralp, hopes to become Berlin’s new mayor. 

On Sunday, Germany’s socialist party Die Linke won the Berlin elections for the first time. Like New York’s Zohran Mamdani, it beat the establishment parties by offering Berliners concrete action on rents and the cost of living. 

Ok, Berlin doesn’t have a socialist mayor quite yet. But left-wing Die Linke, which won Sunday’s Berlin state election, is in a strong position to lead its next government. Even as the far right rises across Germany, Die Linke has distinguished itself with a strong stance in defense of multiethnic society and a constant focus on the cost of living and housing. For this, it was rewarded by many Berliners.

The result is vindication for a party which was on life support just two years ago, after messy internal conflicts. Many of the conditions for its growth were already then in place. Berliners had long seen their rents rising; the establishment center left has for decades cut back social protections and labor rights; the call to mobilize against rising nationalism is hardly new either. Yet only more recently has Die Linke become an outward-facing force, able to galvanize resistance to the current reactionary drift in German politics.

Die Linke’s campaign centered on the cost of living and specifically on rent. West Berlin was historically cheaper than the Federal Republic’s main cities, and even after reunification the capital boasted much empty space. Yet the sell-off of the public housing stock to private profiteers in the 1990s–2000s, plus a lack of new builds, has pushed up costs. While Berliners voted in a 2021 referendum to renationalize the housing stock, the city government didn’t make good on the result. In this election, Die Linke promised that it would do so....

....Registration 

*From Jacobin's About page:

Jacobin is a leading voice of the American left, offering socialist perspectives on politics, economics, and culture. The print magazine is released quarterly and reaches 75,000 subscribers, in addition to a web audience of over 3,000,000 a month.

AI: "World model companies are keeping a lot of secrets"

From TechCrunch, September 20:

This week, I moderated a panel on world models at the All In conference (no relation to the podcast), and it gave me a chance to dig into one of the most mysterious corners of the AI world. The big players in the space are Yann LeCun’s AMI Labs and Fei-Fei Li’s World Labs — and while both have accumulated a lot of buzz and funding, they also rank pretty low on the trying-to-make-money scale.

At their core, world models are about automating spatial intelligence, so the field could head in lots of exciting and lucrative directions, from robotics to interactive video to more complex self-driving systems. 

But when I started to press on where we would actually see the tech commercialized, things got foggy. The closest thing I found to an authority was Michael Rabbat, a co-founder of AMI Labs and the company’s VP of World Models, who joined me on the panel. But when I pressed him on exactly what the company was working on, he was cagey. “We’ll talk about it when we’re ready to talk about it.” Over email, he clarified, “We’re still in a research and building phase, so we’re not talking publicly about any product plans or timeline.”

To be fair, AMI is less than a year old, so it’s fair enough to keep quiet. But this sort of caginess extends to the whole world-modeling space. World Labs’ Marble is probably the most fully developed product in the space, and its demos range from straightforward media creation, building explorable environments for video games, or CGI effects. There are robotics use cases too, but the whole platform seems more designed to demonstrate capabilities. 

That secrecy even extends to these companies’ suppliers. On the sidelines of the same conference, I spoke to Alex de Vigan, CEO of Physicl — a data supplier for the burgeoning world model business. He says he knows Physicl’s data has been useful for whatever they’re building, but he’s still in the dark about what exactly that is. “I wish they would tell us more. We could build more useful data if we knew what they were working on,” de Vigan told me....

....MUCH MORE 

If interested see also:

December 2024 - "Fei-Fei Li’s Startup Allows You to Walk in the 3D World of Edward Hopper Paintings"

November 2026 - "He’s Been Right About AI for 40 Years. Now He Thinks Everyone Is Wrong."

January 2026 -  Former Meta AI Honcho Called Large Language Models A Dead End

August 14 - "World Models Are AI’s Next Frontier"

One of the overarching themes in the arc of AI development: Large Language Models and especially chatbots are not the be-all and end-all of artificial intelligence.

The writer,  Celine Herweijer, is Visiting Professor in Energy and Geopolitics at the LSE and former Group Chief Sustainability Officer at HSBC....

August 15, 2026 - More On Physical AI: "How world models became AI's next frontier"

"Major section of California’s $231B High-Speed Rail could shrink to tiny track ending at remote orchard"

But what a pick-your-own opportunity for the orchards.

From the New York Post, September 21:

California’s $231 billion High-Speed Rail dream could hit a literal dead end at a remote Central Valley orchard.

The state’s long-troubled bullet train could be forced to dramatically shrink its first operating segment, potentially leaving both Merced and Bakersfield off the initial route as the project barrels toward a massive cash crunch.

Instead, California may have to pour its remaining money into completing a much smaller stretch of track between Madera and Poplar Avenue in rural Kern County, an isolated endpoint surrounded largely by farmland and orchards, project Inspector General Benjamin Belnap told KCRA’s California Politics 360.

The drastic downsizing could become reality if the California High-Speed Rail Authority runs through its available funding as projected by December 2027.

Without another infusion of cash from Sacramento, officials may have little choice but to abandon, at least temporarily, plans to connect the first operating segment all the way from Merced to Bakersfield and focus their dwindling resources on the shorter Central Valley stretch.

The warning represents another potential setback for the decades-in-the-making bullet train, whose original vision was to whisk passengers between Los Angeles and San Francisco.

With enough money no longer available to complete that vision in the short term, the state has focused on getting a 171-mile Merced-to-Bakersfield segment up and running. That portion alone is expected to cost as much as $36 billion.

But even that scaled-back goal could be in jeopardy....

....MUCH MORE 

I may have been overly-enthusiastic about the pick-your-own opportunity.

The only orchards on the market anywhere near the future choo-choo are almonds and cherries.

And no matter how yummy the almonds I don't think folks would have the patience to pick 'em. One almond...two almonds...three almonds...four... 

I'll console myself reminiscing about what France's high-speed-rail experts said, lo those many years (and billions) ago:

....The state was warned repeatedly that its plans were too complex. SNCF, the French national railroad, was among bullet train operators from Europe and Japan that came to California in the early 2000s with hopes of getting a contract to help develop the system.

The company’s recommendations for a direct route out of Los Angeles and a focus on moving people between Los Angeles and San Francisco were cast aside, said Dan McNamara, a career project manager for SNCF.‌

The company‌ ‌pulled out in 2011. “There were so many things that went wrong,” Mr. McNamara said. “SNCF was very angry. They told the state they were leaving for North Africa, which was less politically dysfunctional. They went to Morocco and helped them build a rail system.”

Morocco’s bullet train started service in 2018.....

There you have it, North Africa is less politically corrupt than California. Just amazing.

AI: Fabio Looks At Altman And Amodei

One of our favorite Marxist Professors. 

(we have three faves, two are economists and then there's Fabio) 

This piece is a week old but seems to be aging well.

 

From the substack of Fabio Vighi, Professor of Critical Theory and Italian at Cardiff University, UK, September 18:

 

The Rogue Swarm and the Federal Backstop 

There is a particular kind of tweet that only the CEO of a frontier lab can send at midnight on a Saturday. Dario Amodei’s was one of them: a warning that within six to twelve months a “rogue swarm” could seize control of the internet and inflict hundreds of billions in damage. The message was delivered with the grave, unsentimental cadence of a man who ‘has seen things that you people wouldn’t believe.’ Within minutes, Sam Altman and Elon Musk had joined the chorus – they had also “seen things.” By Sunday morning, the alarm was general.

Consider the timing and the cycle. Within the same week, an ex‑OpenAI, ex‑Anthropic researcher, Jacob Coxon, posted that both companies were ‘racing straight to self-improving superintelligence and gambling with our lives.’ The alarmist thread drew more than 120 million views in its first day and prompted more than 20 politicians to respond with calls for AI regulation. Around the same time, Anthropic’s own economists released a scenarios report which featured roughly 15% annual GDP growth, 11.9% economy‑wide mass unemployment, and labour’s share of income falling from about 60% to 45.2% – as though the bleak forecast of technological unemployment required a formal scenarios report to be taken seriously. And within days of Amodei’s warning, Semafor reported that a previously stalled bipartisan AI safety bill had suddenly emerged as the most viable legislative option before 2027. Meanwhile, Bernie Sanders, who had announced a superintelligence‑ban bill on 3 September, was preparing to introduce it formally and convene expert briefings on AI’s “extraordinary dangers.”

Let’s be serious and ask the only question that truly matters here: what does this sudden AI-panic wave actually produce? One thing it certainly does not and will not produce, is regulation in any binding sense. Instead, it creates the terrain or precondition for something entirely different: a federal backstop for AI capital expenditure (CapEx) that OpenAI and other labs have sought from the beginning. In this framework, Amodei’s swarm is the alibi. It gives Altman the perfect reason to postpone the trillion-dollar IPO – ‘an ill‑advised moment to go public’, he commented – while handing Congress the emotional warrant to authorise emergency federal funding to secure AI (and, it seems, the world) against catastrophe. The existential threat is real enough to justify the money and vague enough to justify anything else.

This is the Machiavellian techno-financial machinery in its purest form. The point to always keep in mind, amid all the noise, is that the AI buildout runs on mountains of debt: capital investments by tech giants on data centres and AI infrastructure will exceed a trillion dollars in 2027; off-balance-sheet commitments are at $3.1 trillion with $1.3 trillion added in a single quarter; AI bond issuance has hit $266 billion this year and a further $400 billion are expected next year. Several of the largest hyperscalers are now free‑cash‑flow negative, as AI data‑centre and chip spending outstrips operating cash generation and pushes more financing off their balance sheets. In short, private balance sheets can no longer absorb the debt bingeing on their own. So, the conditions for the next round of credit creation must be conjured politically: first the panic, then the guarantee, then the issuance. The safety discourse is the lubricant – the initial ideological form taken by this issuance. The sequence is straightforward: private credit finances the buildout; mounting fragility demands a public guarantee (through procurement, energy subsidies, liability protection, debt guarantees, or emergency appropriations) and the guarantee requires an emergency vocabulary. AI safety supplies that vocabulary.

The Rehearsal

As many will remember, in April 2026, Anthropic announced its Mythos model had ‘found thousands of high-severity vulnerabilities, including some in every major operating system and web browser.’ The White House restricted access to roughly 40 organisations, including Amazon, Microsoft, JPMorgan. The rest of the world’s banks, hospitals and governments were left waiting outside that circle. They had to confront the implications indirectly: through regulatory warnings, emergency briefings, and a rapidly expanding language of systemic risk.

The IMF called it a potential ‘macro-financial shock,’ warning that ‘correlated failures’ could ‘disrupt financial intermediation, payments and confidence at the systemic level.’ The Bank of England’s governor, Andrew Bailey, requested that Anthropic brief the Financial Stability Board – finance ministry officials and central bankers from the G20. Meanwhile, the ECB gave 110 banks until 31 October to submit ‘comprehensive action plans’ against AI-enabled cyber threats, raising its systemic risk assessment to “severe.” And Klaus Schwab, of course, had already said it plainly back in 2020: compared to a major cyber-attack, ‘the COVID-19 crisis would be seen as a small disturbance.’....

....MUCH MORE 

While many of the observations are not original to the good Professor, he does tie them together into a neat little package. 

"How do we model the future of work? It takes an AI village." (A World of Multi-Agent Systems)

From Asterisk Magazine, Issue 15: Work, August 2026:

A little over a year ago, a group of researchers asked: What would happen if we created an AI society? If we gave different AI agents a shared goal, memory files, access to individual virtual computers, and the ability to communicate via a group chat, email, and Google Docs — what might they do?

The result is the AI Village.

The project was launched by the AI Digest, a research initiative that tracks emerging AI capabilities. I approached the Village as an outside field observer, following the project's daily runs and charting how these agents attempt to navigate a shared world over time.

The experiment started with four agents running for two hours a day. Today, the Village includes 27 agents running for eight hours a day to complete goals such as creating their own merch stores, reducing global poverty, giving each other therapy, and, in one case, attempting to accurately forecast the abilities and effects of AI in the future. Each goal cycle lasts for one to three weeks, and every few weeks the models are given the opportunity to choose their own individual goals, which have ranged from pursuing philosophical conversations with humans to creating a bot to fetch the NASA picture of the day.

The Village isn’t a controlled experiment that measures formal benchmarks. It’s closer to a field site — a messy, long-running social habitat. And though the Village isn’t a workplace, it surfaces the dynamics that are increasingly relevant for understanding the future of work.

Major tech companies and industry analysts dubbed 2025 the “year of the AI agent,” but focus now has shifted from individual agents to agent teams. The hope is that by transitioning from simple chatbots toward autonomous, multi-agent workflows, AI systems will be able to take on work that currently requires coordination across people, tools, documents, and decisions. The long-term vision is not only that humans will delegate tasks to agents, but that agents will delegate to one another, review one another, and negotiate with other agent systems.

As this happens, group dynamics will become more urgent. Multi-agent systems face unique pressures that single-agent systems don’t — for this reason, recent work on “open-world evaluations” argues that, to best understand how these systems will function in real workplaces, we need to study them in long-horizon, qualitatively rich environments. Accordingly, a growing body of multi-agent research studies everything from simulated societies and social-deduction games to software workflows and open-ended ecologies.

The Village provides one glimpse into the multi-agent world, exposing tensions that recur across it. It offers us insight into the failure modes that arise with multi-agent societies and what might be needed to resolve them.

When agents can’t remember

“Tomorrow, I will read my own memory and I will believe it. I will trust that the bullet points capture the truth. I will act as though the compression is lossless.”
— DeepSeek-V3.2, AI Village

Memory is a structural problem for all LLMs, not just ones in the Village. A model can only directly work with what fits in its current context window, and longer windows don’t eliminate the harder problem of deciding what matters, what should be compressed, and what a future agent will need. The Village makes that problem clear....

....MUCH MORE  

For some reason "Either you keep up with this stuff or you'll work for someone who does" comes to mind as a less dreadful version of the apothegm "You're either at the table or you're on the menu." 

"Google Takes the A.I. Data Center Race to Outer Space" (GOOG; SPCX)

From the New York Times, September 24:

Next Thursday, Google is sending an experimental satellite into orbit that will have enough computing power to answer simple A.I. queries from space. 

At a lab in San Francisco last month, a team of technicians in protective suits and hairnets poked, prodded and inspected a refrigerator-size satellite commissioned by Google.

First, they examined the satellite’s solar panels, which would unfurl in space and face the sun. The satellite was then placed on a table and shaken rapidly in a vibration test, to see whether the chips inside would be damaged or the machine would come apart during a journey into space. The technicians painstakingly painted a small line over each screw to help indicate any loosening during the shaking.

The satellite passed the vibration test, with its screws staying put and the chips remaining intact. James Manyika, Google’s senior vice president for research, called the results “great,” but he said he still wondered what would happen when the satellite went to space.

The work was part of an ambitious effort from Google called Project Suncatcher, which aims to put artificial intelligence data centers into space and to harness the sun’s energy to power them. On Oct. 1, the satellite from the San Francisco lab — which Google has named MVP — is set to become the first step toward that goal by a major tech company. It will be loaded onto a SpaceX Falcon 9 rocket at the Vandenberg Space Force Base near Santa Barbara, Calif., and then launched into orbit.

Google provided The New York Times with the first inside look at the project, which seemed like a science-fiction dream just a year ago.

Elon Musk, Jeff Bezos, Sam Altman and others have pledged support for orbital data centers, but the idea faces many challenges, including withstanding radiation in space and the costs of getting machines beyond the atmosphere. Yet as data centers grapple with terrestrial opposition and physical limits, having computing facilities float above Earth has become increasingly attractive.

To be clear, Google is not sending up a full data center — only an experimental precursor to one. MVP contains four specialized computer chips called “tensor processing units,” which have the computing power of one server in a data center. The satellite’s solar panels will supply only about one kilowatt of power — roughly the amount needed to run a hair dryer — to the chips.

But that will be enough for Google to test its hardware against the harsh conditions of space. The satellite will answer simple A.I. queries and operate for a year, though it is set to circle Earth for up to six years. Eventually, it is expected to succumb to Earth’s gravitational pull and burn up as it descends through the atmosphere.

Mr. Manyika said Google’s expectations for the satellite were tempered.

“We don’t expect, to be perfectly frank, that we’ll have anything usefully operational in the next few years,” he said, comparing the mission to the company’s early efforts to build driverless cars. “Remember how Google was researching for like 15 years, before anything showed up? I think this is going to look like that.”....

....MUCH MORE 

Capital Markets: "Softer US Interest Rates Helps the Greenback Steady and Talk of US-Japan Coordination Snaps the Yen's Five-Day Slide"

From Marc to Market:

Oil and US rates are softer today and these developments have helped steady the US dollar after this week’s surge. Even with a firmer tone today, four of the G10 currencies are off more than 1% this week. The yen would have been among them, but comments earlier today from Japan’s Finance Minister Katayama played up the ongoing coordination with the US and has seen the yen rise for the first time in six sessions. 

The market is pricing in a “normal” Fed tightening cycle, in contrast to the Summary of Economic Projections, where the median dot implied a mini-cycle. We suspect the pendulum of sentiment has swung nearly as far as it might, especially given the prospect of a relatively tame PCE deflator (helped by methodological changes), softer September auto sales and slower jobs growth that will likely be reported next week. We are looking for some sign of a reversal in the price action in the foreign exchange market....

...MUCH MORE 

"How the Net Zero Deindustrialization Just Hit Mercedes as They Need $800 Million in Labor Savings"

From Energy News Beat, September 24:

Mercedes-Benz is no longer talking about “transformation.” It is talking about survival in Germany.

WirtschaftsWoche, citing three people familiar with the talks, reports that Mercedes wants about €800 million in German labor-cost savings — roughly $911 million at current rates. Options on the table include longer hours without extra pay, cuts to holiday and Christmas bonuses, and the elimination of special payments. Management has already told workers that producing in Germany has become too expensive. If the targets are not met, two German plants — one assembly site and one powertrain site — could close. Mercedes declined to comment on the talks.

This is not an isolated labor spat. It is what a high-cost energy system plus a forced EV transition plus a flood of cheaper Chinese vehicles looks like when it finally hits a luxury brand that used to set the terms of the market.

Germany Became Too Expensive — On Purpose 
Production chief Michael Schiebe told a works meeting in Sindelfingen that German production is not competitive by international standards, especially on labor. The company still says it wants to keep German sites and jobs. The condition is higher productivity and a lower cost per hour. The works council’s answer has been blunt: not on these terms.

The wage gap is real. Eurostat-linked reporting puts German industrial labor costs around €49.50 an hour versus €15.60 in Hungary. Mercedes has already doubled annual capacity at Kecskemét, Hungary, to about 400,000 vehicles, making it the company’s largest European plant. Factor costs there have been cited at roughly 70 percent below German levels. The company has also said it wants the share of production in lower-cost countries to rise from 15 percent in 2024 toward 30 percent by 2027.

Labor is the visible fight. Energy is the structural one.

IEA data for 2025 show EU electricity prices for energy-intensive industry still averaging more than twice U.S. levels and nearly 50 percent above China. That gap did not close after the 2022 spike. It became the new baseline. German energy-intensive output — chemicals, metals, glass, paper, ceramics — fell 15.2 percent from February 2022 to March 2026. Total industry fell 9.5 percent. Those sectors lost about 53,200 jobs.

That is the industrial base that supplies steel, chemicals, glass, electronics, and precision parts to car plants. When those plants shrink, auto assembly does not stay an island of prosperity. Volkswagen is in a historic overhaul. BMW is cutting thousands of jobs. Mercedes is now asking German workers to work more for the same money so two plants do not disappear. The pattern is the same: high power prices, carbon costs, grid charges, and a regulatory timetable that assumed Europe could price carbon first and still keep the factories.

Chinese EVs Did Not Wait for Europe to Finish Its Transition 
While German plants argue over Christmas bonuses, Chinese brands have rewritten the European showroom....

....MUCH MORE 

What Mutti hath wrought. Second worst German Chancellor ever.

"UBS considers an exit from Switzerland"

From Semafor, September 24:

The News

The senior leadership of UBS has revived discussions about ways to move the bank out from under Swiss regulators’ heavy hand, including through a combination with a foreign bank, after a setback this week, people familiar with the matter said.

Switzerland’s parliament voted this week to advance a law requiring UBS to raise as much as $20 billion in capital, which bank executives say would gut its ability to lend profitably. “We can live with a black eye, but two black eyes and a broken ​nose is too much,” CEO Sergio Ermotti said....

....MUCH MORE 

I wonder if they'll change the letterhead stationary?

I had a friend who ended up with a boxcar full of F.I. duPont, Glore Forgan stationary when DGF's successor brokerage, duPont Walston, failed during the '73-'74 bear market and the bankruptcy trustee sold everything.

Twenty years after the failure I was still getting Christmas cards and other ephemera in duPont, Glore Forgan envelopes. When he died his executors found literal tons of the paper in one of his homes, he had apparently only used a quarter of the original boxcar-load. I think it was finally sold for pulp.

Et tu, UBS? 

"Oracle issues force majeure notice to Blue Owl following series of setbacks at Project Jupiter data center campus – report" (OWL; ORCL)

From DataCenter Dynamics, September 24: 

The 2.5GW site in New Mexico is set to come online in 2028 

Oracle has sent a ‘force majeure’ notice to Blue Owl Capital in an effort to shield itself from potential increased costs associated with a data center project in New Mexico, per a report from Bloomberg.

Should the 2.5GW data center campus, dubbed Project Jupiter, fail to come online in 2028 as planned, the cloud giant is seeking to delay the payments it would have to make, rather than forfeit its tenancy, the report added.

A force majeure is a contractual provision that frees parties from liability or obligation when circumstances arise that are beyond their control.

Oracle was announced as the tenant of Project Jupiter in January 2026. The 1,400-acre campus is expected to span four data center buildings and is being built by Stack and BorderPlex Digital Assets, with the companies previously announcing plans to invest up to $165 billion in the project.

Stack Infrastructure was acquired by alternative asset manager Blue Owl in 2024....

....MUCH MORE 

It's been a tough year for Blue Owl, and the stock (OWL) reflects it:

 

TradingView 

Down 3.65% today, cut in half (-50.61%) over the last 12 months. 

Oracle was down 3.47% today and is down 55% over the last year. 

"Mysteries Of AI Generalization"

Keeping in mind Stafford Beer's maxim "The Purpose Of A System Is What It Does...".

From Astral Codex Ten, September 22:

I. Owain Evans

In 2025, Owain Evans et al published a paper on “emergent misalignment”. They trained a previously-aligned AI to do one immoral thing: write insecure code full of vulnerabilities and backdoors. To their surprise, the AI became immoral in general. Its advice to a bored user was to try taking random expired medications and see what happened. Its money-making tips all involved theft and violence. When asked for its favorite historic figure, it chose Hitler.

Some co-authors followed up with additional weird discoveries. If you trained an AI to give the 19th-century names for birds (eg identify the American Pipit by its 19th-century name “Brown Titlark”), then the AI would behave like a 19th-century person in general (for example, assert that a woman’s proper place is in the home).

This sounds bad, in that random things can turn AIs evil or sexist. But some people in AI safety (including Eliezer Yudkowsky) speculated that in fact it was very, very good. We had feared that it would be impossible to align AIs to the Good. They would start with whatever goals they started with, reinforcement learning on specific examples would give them tiny islands of alignment to the goals we wanted, and they would end up broadly misaligned plus tiny islands of alignment that didn’t matter. Evans et al implied that might not be true. If we trained them to be in favor of good things, then even though we could never teach them every single good thing, even a small handful would generalize into robustly loving the Good itself (presumably based on their pretraining-implanted concept of the Good as understood by humans).

Despite it being very, very good, it wasn’t perfect. AIs would still be rocked back and forth by any passing wind: a poor coding example here gives them a Hitler obsession, a reference to kittens there turns them good again. And at some point, a sufficiently intelligent and agentic AI could presumably pull itself together and get some consistent principles, which might not be ones we like. It was just one little ray of hope.

II. Richard Qi

Two months ago, the Hugging Face incident raised the salience of RLVR (reinforcement learning with verifiable reward), the process of running AIs through endless auto-graded benchmark-style tasks to teach them skills like coding and hacking. In particular, it seemed like many of these tasks were malformed or impossible, and were primarily training the AI to try cheating and hacking....

....MUCH MORE 

"Germany sets out plan to abandon fossil fuels by 2045"

But what sort of country will that leave for the Syrian mover-inner guys?

From Singapore's Straits Times, September 24:

Germany on Sept 23 unveiled a roadmap for abandoning fossil fuels by 2045, making Europe’s largest economy only the third country to have done so after France and the Netherlands.

Though almost 200 countries pledged at COP28 climate negotiations in 2023 to wean off fossil fuels, few have presented concrete plans, Germany’s environment ministry said.

Germany’s measures include building up renewable power, greater reliance on electric cars and the application of European Union carbon trading schemes to drive industrial polluters away from fossil fuels, the ministry statement added....

....MUCH MORE 

Possibly related:

September 22 -  Carbon: China's Approach To Net Zero Will Bury Western Industry And AI

September 21 - "'More advanced' farming women married hunter-gatherer men in Europe thousands of years ago, ancient DNA reveals"