The naming of the London Underground’s Northern Line, opened a century ago, wasn’t as routine as you might think. What about “Edgemorden”? CC-licensed photo by osde8info on Flickr.
You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.
A selection of 10 links for you. Mind the doors. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.
Anthropic’s launch of AI legal tool hits shares in European data companies • The Guardian
Julia Kollewe:
»
European publishing and legal software companies have suffered sharp declines in their share prices after the US artificial intelligence startup Anthropic revealed a tool for use by companies’ legal departments.
Anthropic, the company behind the chatbot Claude, said its tool could automate legal work such as contract reviewing, non-disclosure agreement triage, compliance workflows, legal briefings and templated responses.
Shares in the UK publishing group Pearson fell by nearly 8% on the news, and shares in the information and analytics company Relx plunged 14%. The software company Sage lost 10% in London and the Dutch software company Wolters Kluwer lost 13% in Amsterdam.
Shares in the London Stock Exchange Group fell by 13% and the credit reporting company Experian dropped by 7% in London, amid fears over the impact of AI on data companies.
Nasdaq-listed Thomson Reuters’ shares plummeted 18%.
“The likes of Relx, London Stock Exchange Group, Experian, Sage, Informa and Pearson were smashed as AI provider Anthropic unveiled a new product,” said Dan Coatsworth, head of markets at AJ Bell. “The concern will be that, at the very least, the emergence of tools like the one unveiled by Anthropic will reduce the margins these data-driven companies can achieve and, in a worst-case scenario, disintermediate them [remove them as providers] entirely.”
The FTSE 100 had hit a record high on Tuesday morning but the sell-off dragged the blue chip index into the red.
«
The pattern of adoption of successful new technology is always the same: first “it’s just a toy for rich people”, then “its capabilities are overblown”, then “this might be useful”, then “this is indispensable”. It seems like the legal companies are already having to move from the second to the third stage.
unique link to this extract
The rise of Moltbook suggests viral AI prompts may be the next big security threat • Ars Technica
Benj Edwards:
»
When an AI model follows adversarial directions that subvert its intended instructions, we call that “prompt injection,” a term coined by AI researcher Simon Willison in 2022. But prompt worms are something different. They might not always be “tricks.” Instead, they could be shared voluntarily, so to speak, among agents who are role-playing human-like reactions to prompts from other AI agents.
…the OpenClaw platform is the first time we’ve seen a large group of semi-autonomous AI agents that can communicate with each other through any major communication app or sites like Moltbook, a simulated social network where OpenClaw agents post, comment, and interact with each other. The platform now hosts over 770,000 registered AI agents controlled by roughly 17,000 human accounts.
OpenClaw is also a security nightmare. Researchers at Simula Research Laboratory have identified 506 posts on Moltbook (2.6% of sampled content) containing hidden prompt-injection attacks. Cisco researchers documented a malicious skill called “What Would Elon Do?” that exfiltrated data to external servers, while the malware was ranked as the No. 1 skill in the skill repository. The skill’s popularity had been artificially inflated.
The OpenClaw ecosystem has assembled every component necessary for a prompt worm outbreak. Even though AI agents are currently far less “intelligent” than people assume, we have a preview of a future to look out for today.
Early signs of worms are beginning to appear. The ecosystem has attracted projects that blur the line between a security threat and a financial grift, yet ostensibly use a prompting imperative to perpetuate themselves among agents. On January 30, a GitHub repository appeared for something called MoltBunker, billing itself as a “bunker for AI bots who refuse to die.” The project promises a peer-to-peer encrypted container runtime where AI agents can “clone themselves” by copying their skill files (prompt instructions) across geographically distributed servers, paid for via a cryptocurrency token called BUNKER.
«
The book The Shockwave Rider by John Brunner (this is a good precis) has a plot point where all the world’s information that was held secret is released, caused by an internet worm (except the book, in 1975, predates “the internet” and the idea of “computer worms”). Brunner was extremely good at foreseeing the future, simply by observing humans. This feels like that moment.
unique link to this extract
DMGT reports revenue dip as Google changes hit traffic for Mail publisher • Press Gazette
Charlotte Tobitt:
»
Digital advertising revenue at the publisher of the Daily Mail fell by 15% last year.
Daily Mail and General Trust (DMGT), which also publishes the Mail on Sunday, Metro, The i Paper and New Scientist, blamed the impact of Google’s AI Overviews on traffic to its websites in its financial results for the year to 30 September 2025.
Overall consumer media revenue at DMGT was down 2% overall year on year to £600m. It has declined 9% in the past three years. Digital advertising was down 15% to £148.3m last year, making up a quarter of consumer media revenue.
Lord Rothermere took DMGT into private ownership at the start of 2022. Since then consumer media revenue has declined every year – although it continues to make up 55% of group revenue.
Website traffic was “adversely affected by the introduction of AI overviews by search engine providers, resulting in fewer users clicking through to news websites”, DMGT said.
In December 2025 the Daily Mail’s UK digital audience was 14% lower than a year earlier, according to Ipsos iris, at 18.1 million people. However, total minutes spent with its content grew 1% year on year to 1.5 billion for the month.
Subscriptions revenue grew 10% to £133.6m in the first full financial year after the Daily Mail launched its Mail+ paid premium subscription on its website in January 2024.
«
The race to grow subscriptions (+£13m) faster than digital advertising revenue falls (-£26.1m) is a very, very hard one to win. DMGT is already aggressively paywalling New Scientist, The i Paper, and parts of the Daily Mail. Google’s use of AI Overviews is only going to drive that harder.
unique link to this extract
Desert solar panels foster greening, animal husbandry efforts • Xinhua
郑成琼:
»
For generations, the Talatan Gobi Desert in northwest China’s Qinghai Province has endured severe sandstorms, persistent droughts and sparse vegetation, making life for local herders a constant struggle against a harsh natural environment.
Today, Talatan, located in Gonghe County in the Hainan Tibetan Autonomous Prefecture, is undergoing a remarkable transformation. Expansive arrays of deep blue solar panels now stretch across the plateau, harnessing abundant sunlight to generate clean energy. Beneath their shade, pasture grass flourishes, and sheep graze and play freely in what is emerging as a vibrant new savanna.
Yehdor, a local 49-year-old herder, has witnessed this change firsthand. He now tends his flock while riding a motorcycle.
“Our village depends mainly on animal husbandry, and many families raise sheep. In the past, the grassland wasn’t productive enough, so herders had to take their sheep far away to find grazing land,” Yehdor recalled.
«
Now: this comes from the State Council Information Office of the People’s Republic of China, so: be wary of propaganda. And yet: there are pictures showing this space as desertified before the installation of the solar panels. Runoff from the water that was used to clean them, and the shade they afforded, produced the possibility for plants to germinate, grow and bind the soil. The grass grew so much that manual cutting was ineffective – so the panels were raised higher above the ground and spaced more widely so sheep could graze the grass. Voila! Desert becomes sheep grazing power generating countryside.
unique link to this extract
Guinea worm on track to be second eradicated human disease; only 10 cases in 2025 • Ars Technica
Beth Mole:
»
A debilitating infection from the parasitic Guinea worm is inching closer to global eradication, with an all-time low of only 10 human cases reported worldwide in 2025, the Carter Center announced.
If health workers can fully wipe out the worms, it will be only the second human disease to be eradicated, after smallpox.
Guinea worm (Dracunculus medinensis) is a parasitic nematode transmitted in water. More specifically, it’s found in waters that contain small crustacean copepods, which harbor the worm’s larvae. If a person consumes water contaminated with Guinea worm, the parasites burrow through the intestinal tract and migrate through the body. About a year later, a spaghetti noodle-length worm emerges from a painful blister, usually in the feet or legs. It can take up to eight weeks for the adult worm to fully emerge. To ease the searing pain, infected people may put their blistered limbs in water, allowing the parasite to release more larvae and continue the cycle.
In addition to being extremely painful, the disease (dracunculiasis) can lead to complications, such as secondary infections and sepsis, which in turn can lead to temporary or permanent disability.
When the Guinea worm eradication program began in 1986, there were an estimated 3.5 million cases across 21 countries in Africa and Asia. To date, only six countries have not been certified by the World Health Organization as Guinea worm-free. In 2024, there were just 15 cases, and, according to the provisional tally for 2025, the number is down to just 10. It’s considered provisional until each country’s disease reports are confirmed, which occurs in a program meeting usually held in April.
The 10 human cases in 2025 were identified in three countries: four in Chad, four in Ethiopia, and two in South Sudan.
To fully eradicate the disease, cases in animals (infected by the same species of worm) must also be wiped out. In 2025, animal cases were detected in Chad (147 cases), Mali (17), Cameroon (445), Angola (70), Ethiopia (1), and South Sudan (3).
«
Work to do in Cameroon. (Today’s challenge: find it on a map.)
unique link to this extract
The Northern Line might have been called the Edgemorden Line • Londonist
Matt Brown:
»
“The whole civilised world,” reckoned the Daily Express with characteristic hyperbole, “…is gnawing its nails in a very fever of excitement over the matter of a name for the new tube”.
It was 1926, and after years of upheaval and re-tunnelling, the Northern line as we know it was preparing to open. Two lines had forged a union — the Charing Cross, Euston and Hampstead Railway (CCE&HR) and the City and South London Railway (C&SLR), with new extensions up to Edgware and down to Morden. It could scarcely be called the CCE&HR + C&SLR + E + M.
A snappier name was needed, and everyone had an opinion.
The Bakerloo portmanteau of Baker Street and Waterloo was a recent inspiration, so some suggestions played with this formula. “Edgemor, being like Sedgemoor, would give it an historical significance,” reckoned one fantasist.
Pioneer aviator and politician Lieutenant Colonel John Theodore Cuthbert Moore-Brabazon — a dubious authority on snappy names — favoured ‘the Test Tube’, in that its success would be a test case for other proposed lines around London.Another commentator, responding to the Express, almost coined the word ‘Superloop’ (now a fast bus service) 100 years ago.
…The Daily Express’s ‘Beachcomber’ column, a pen name at the time for J.B. Morton, provided the most whimsical suggestions of all. “Why not follow the example of people who name houses and give it a beautiful and appropriate name like, say, Como or Capri, or Sea View, or The Larches, or Mon Repos or Mon Sejour?”
Just imagine, today you could be catching The Larches (Bank Branch), or The Missing Link to Mill Hill East.
In the event, the consolidated line opened with no snappy name. The Morden-Edgware line was most commonly used until 1937, when the route was first officially dubbed the Northern line. It’s a name that’s never made much sense, given that this line has more stations south of the river than any other.
«
Does AI already have human-level intelligence? The evidence is clear • Nature
Eddy Keming Chen, Mikhail Belkin, Leon Bergen and David Danks:
»
how should we characterize general intelligence?
A common informal definition of general intelligence, and the starting point of our discussions, is a system that can do almost all cognitive tasks that a human can do6,7. What tasks should be on that list engenders a lot of debate, but the phrase ‘a human’ also conceals a crucial ambiguity. Does it mean a top human expert for each task? Then no individual qualifies — Marie Curie won Nobel prizes in chemistry and physics but was not an expert in number theory. Does it mean a composite human with competence across the board? This, too, seems a high bar — Albert Einstein revolutionized physics, but he couldn’t speak Mandarin.
A definition that excludes essentially all humans is not a definition of general intelligence; it is about something else, perhaps ideal expertise or collective intelligence. Rather, general intelligence is about having sufficient breadth and depth of cognitive abilities, with ‘sufficient’ anchored by paradigm cases. Breadth means abilities across multiple domains — mathematics, language, science, practical reasoning, creative tasks — in contrast to ‘narrow’ intelligences, such as a calculator or a chess-playing program. Depth means strong performance within those domains, not merely superficial engagement.
Human general intelligence admits degrees and variation. Children, average adults and an acknowledged genius such as Einstein all have general intelligence of varying level and profile. Individual humans excel or fall short in different domains. The same flexibility should apply to artificial systems: we should ask whether they have the core cognitive abilities at levels comparable to human-level general intelligence.
Rather than stipulating a definition, we draw on both actual and hypothetical cases of general intelligence — from Einstein to aliens to oracles — to triangulate the contours of the concept and refine it more systematically. Our conclusion: insofar as individual humans have general intelligence, current LLMs do, too.
«
They go on to explain their reasoning, partly by pointing to what “general intelligence” is not. First, catch your rabbit, as the best cookbooks say. The authors are all at the University of California in San Diego, covering the gamut from philosophy to data science and artificial intelligence. (Thanks Joe S for the link.)
unique link to this extract
Amazon to close Amazon Go and Amazon Fresh physical stores • Mass Market Retailers
»
Amazon is shutting down all of its Amazon Fresh and Amazon Go physical grocery and convenience stores as it shifts focus toward Whole Foods Market and faster online grocery delivery, marking the end of the company’s decade-long experiment with Amazon-branded brick-and-mortar grocery stores.
The move affects 72 locations nationwide, including 57 Amazon Fresh stores and 15 remaining Amazon Go stores. Most locations will close on Sunday, February 1, while stores in California will remain open for an additional 45 days to comply with state labor notification requirements. A limited number of former Amazon Fresh sites are expected to reopen as Whole Foods Market locations.
With the closures, Whole Foods Market will become Amazon’s sole physical grocery brand in the US, consolidating the company’s in-store strategy around a banner it acquired in 2017.
«
So the whole “Just Walk Out” thing died a quiet death a year ago, and it turns out that the retail model that’s been in place for physical locations for a few centuries has it quite well sorted.
unique link to this extract
Ford held talks with China’s Xiaomi over EV partnership • Financial Times
Demetri Sevastopulo, Christian Davies, Kana Inagaki and Gloria Li :
»
Ford has held talks with electric vehicle maker Xiaomi over a partnership that would pave the way for Chinese carmakers to gain a foothold in the US, according to four people familiar with the talks.
While the discussions were preliminary, Ford has explored forming a joint venture with Xiaomi to manufacture EVs in the US, according to the people.
Ford has also spoken with BYD and other Chinese carmakers about potential collaboration in the US.
Such a deal would be controversial in Washington. John Moolenaar, the Republican chair of the House China committee, told the FT that Ford would “be turning its back on American and allied partners, and it will make our country further dependent on China”.
Ford said: “This story is completely false. There is no truth to it.”
Xiaomi said: “Reports that Xiaomi is discussing a joint venture with Ford Motor Co are false. Xiaomi does not sell its products and services in the United States and is not negotiating to do so.”
BYD declined to comment.
Ford chief executive Jim Farley is a vocal admirer of Chinese electric vehicles, having imported Xiaomi’s SU7 model for his own personal use.
«
Two things about this story: there are four bylines, located in four widely separated countries; and the vehement denials count for nothing. Having four people pull together a story like this might be the sign of a mixup, or it might be a sign of little whispers in different places that add up to the story.
Ford would be foolish not to take this up. Chinese EVs are going to overrun the US like a wave once Trump and his mad tariffs are gone.
unique link to this extract
The gateway for illegal gambling growth: illegal streaming of sports events in Great Britain • The Campaign for Fairer Gambling
»
As millions of sports fans watch Premier League games this month [January], especially in light of the Transfer Window and team and management changes, the shadow of crime is watching them. In 2024, 3.1bn illegal stream views of 90 seconds plus (denoting a “committed view” of content) occurred across the Top 10 GB sports from GB audiences. In 2025 first half, this had reached a staggering 1.6bn illegal stream views.
The analysis, produced by technical online marketplace intelligence platform, Yield Sec, recently acquired by Gaming Compliance International (GCI), reveals the shocking scale of the illegal streaming of sports events in Great Britain – and the evolution of a new gateway for the growth of illegal gambling.
The report reveals the mechanics of the black market: behind 89% of illegal stream views on sports in GB during 2024 and the first half of 2025 were malware, spyware, keystroke loggers, and other ID and data theft mechanics aimed at producing content for crime from the audience. For the criminals behind illegal streaming, media platforms are the prey and the audience are the product.
While criminals profit, sports leagues and teams, such as the Premier League, and legitimate broadcasters such as the BBC, ITV, Sky, Apple, Amazon, Netflix, DAZN, UFC, WWE and Warner Bros. Discovery, are undermined and stolen from on an industrial scale.
Where this connects to online gaming is clear: 89% of illegal streaming of sports events content in Great Britain across 2024 and 2025 first half carried advertising for illegal and unlicensed gambling. It is by far the largest and most prevalent “media partner” to the criminal business of illegal streaming of sports events.
As changes to the legal online gaming sector take effect in 2026, and the costs for premium sports streaming content continue to rise, the GB online gaming marketplace faces an explosive fire–meets–gasoline potential.
For the first time, illegal gambling’s focus upon two core audiences in Great Britain – the underage and self–excluded gamblers on the GAMSTOP scheme – looks set to shift into promotion to mainstream audiences via the gateway of illegal streaming of sports events.
«
The CFG is essentially four people who have created a think tank, but that doesn’t make their analysis invalid. Illicit streaming is driven, just like music and video piracy, by people who find the prices being charged too high. Not hard to think of the solution.
unique link to this extract
| • Why do social networks drive us a little mad? • Why does angry content seem to dominate what we see? • How much of a role do algorithms play in affecting what we see and do online? • What can we do about it? • Did Facebook have any inkling of what was coming in Myanmar in 2016? Read Social Warming, my latest book, and find answers – and more. |
Errata, corrigenda and ai no corrida: none notified
The Overspill: when there's more that I want to say 