If you go searching for a pharmacy in London on Google Maps, you might get more exotic drugs than you bargained for. CC-licensed photo by Ged Carroll 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 9 links for you. Totally wired. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.
Amazon blocks Meta’s Muse AI assistant in new standoff over agentic shopping • GeekWire
Todd Bishop:
»
Amazon says it has cut off Meta’s new Muse personal AI agent from shopping on Amazon.com on behalf of customers, after attempting unsuccessfully to get the Facebook parent company to voluntarily exclude the e-commerce site from the experience.
The problem, Amazon says, is that it never agreed to any of it. Meta didn’t tell Amazon that Muse would access its store, the agent doesn’t identify itself when it browses, and it appears to capture and store customer credentials, which the company says could create privacy and security risks.
As of Sunday night, people trying to use Muse to shop on Amazon were seeing the popup, “Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed.”
“We think it’s fairly straightforward that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service provider decisions about whether or not to participate,” an Amazon spokesperson said in a statement.
Meta did not immediately respond to a request for comment Sunday night.
The company said previously that Muse “has no visibility into people’s passwords or payment methods,” and that credentials a user shares “go into secure storage, so Muse can use them without seeing them, including passwords a person types into the browser themselves.”
The standoff between the tech giants is especially notable because the two companies work together. Amazon products have been purchasable inside Facebook and Instagram since 2023, and Meta signed a multibillion-dollar deal in April to run agentic AI workloads on Amazon’s Graviton chips.
It’s part of a larger debate over who controls the online shopping experience and the customer relationship when AI agents buy items on behalf of consumers.
«
And so the war between the machines begins.
unique link to this extract
Stop relying on chatbots for customer care, UK service providers urged • The Guardian
Robert Booth:
»
Stop relying on AI chatbots for customer care and guarantee the “right to talk to a human”, Citizens Advice has urged essential service providers, as research found they wasted time, caused stress and delayed problem solving for more than half of users.
The spread of the AI-powered systems to provide help about the provision of vital services such as energy, banking, phones and internet, is making it harder for millions of already digitally excluded people to handle snags, according to the frontline charity that last year provided more than 2.7 million people with one-on-one help in England and Wales.
The charity concluded that chatbots caused harm, saying too often they trapped humans in an “endless loop” in which nothing got resolved, and almost half of people who had to use a chatbot found them unhelpful, with some concluding the chatbot seemed rigged to “make me give up”.
It called for service providers to respect a new right for customers to speak to a human after it investigated the wider impact of digital exclusion, estimated to affect at least 14 million UK citizens. It uncovered slower and less reliable support for people who struggle with or have limited access to online systems and found the erosion of phone, face-to-face and paper communication had created a two-tier support system.
“Going round in circles with a chatbot isn’t just annoying, it can be a blockade to essential services,” said Tom MacInnes, policy director at Citizens Advice. “We’ve seen organisations’ overreliance on chatbots and online forms leaving people unable to fix billing problems, sort out their debt or even access homelessness prevention services. There’s a role for chatbots, but the quiet erosion of face-to-face and telephone services is letting too many people fall through the cracks.”
That erosion of traditional ways of communicating means that people without good digital access can face serious difficulties getting simple problems fixed. For example, the charity found that submitting energy meter readings to correct a wrong bill can take less than an hour online if systems are working properly. Without good digital access it takes as long as six weeks offline, putting people “at risk of further harm”.
«
Used to be humans operating off a script from which they couldn’t deviate was the problem; now it’s chatbots.
unique link to this extract
Usenet-Rewind • Erie Data Systems
Erie Data Systems:
»
Search Usenet newsgroup archives — text posts from 1981 to present.
Usenet-Rewind is a research archive for the newsgroup conversations that predate the modern web — early tech support and software debates, scientific and academic discussion, hobbyist and fan communities, entertainment and news commentary, and firsthand accounts of internet history as it happened on Usenet.«
Usenet has had a storied but also complicated history. It’s essentially a set of forums with public messages passed among servers around the world. It was social media before social media, and instituted a lot of the behaviour (good and certainly bad) that one sees today.
Then the explosion of the internet following the dot-com boom meant people had other distractions, and Usenet wasn’t an advertising forum so it really wasn’t of interest to big internet companies or ISPs; more a potential problem for legal reasons. Deja News sucked up the archives and stuck advertising on it; then Google bought it. Then Google lost interest, of course, in February 2024. So now you have this, bringing it back to a sort of life.
unique link to this extract
This AI agent has written thousands of Alaska news articles. We interviewed it • Anchorage Press
Nathaniel Herz:
»
In the past 24 hours, the website AlaskaNews.com has featured stories on city spending on housing for people experiencing homelessness in the Western Alaska hub town of Bethel; on limits on crab harvests in the Bering Sea; and on a proposed underwater data center off the Kenai Peninsula.
Since its inception earlier this year, Alaska News says it’s published more than 4,500 stories.
But the site has no flesh-and-blood reporters. Overseen by a human editor, Cale Green, Alaska News instead employs a team of AI agents that scoop up public documents, videos and announcements and uses them as a basis for stories.
The project is the brainchild of five Alaskans.
Its chief technologists are twins Lee and Lucas Brown — veteran coders and startup entrepreneurs who live in Valdez, make music with AI and call themselves the “Geeks in the Woods.”
Also on Alaska News’ masthead are Matt Larkin, a longtime conservative Alaska pollster, political operative and entrepreneur; Grant Robinson, a former television reporter whose current day job is working as a spokesperson for Republican Gov. Mike Dunleavy; and Green, a videographer and former part-owner of the Alaska Landmine news site who’s also done campaign work for centrist candidates and causes.
Cale Green, center, and twins Lee and Lucas Brown lead AlaskaNews.com. (Nathaniel Herz/Anchorage Press)
“We don’t impose judgment on the things we’re reporting,” said Lee Brown, a registered independent. “We try to provide both sides.”Nonetheless, the site, and the political track records of its founders, raise profound questions about journalism, editorial judgment and the future of the media business.
While some stories, particularly those related to campaigns, are flagged for review by Green, the editor, before publication, other pieces are published without any direct human involvement and scrutinized by Green only afterward — a departure from the standards applied by mainstream newsrooms that treat AI-generated material as unvetted.
After sitting down for an hour with its human founders, we wanted to hear directly from the computer programs that do much of Alaska News’ legwork. So, in an upstairs office in Green’s home, he set up Anchorage Press Executive Editor Nathaniel Herz to “interview” Walter, the AI agent that’s written most of Alaska News’ stories.
«
The interview is fascinating; the chatbot (OpenClaw) becomes elusive when quizzed sufficiently hard, and never really commits to anything. But it never gets bored by stories that would have humans screaming for release.
unique link to this extract
GBNames • Geographic Data Service
JT van Dijk:
»
Our research, carried out at the ESRC Consumer Data Research Centre (GeoDS), is exploring the generational and inter-generational residential movements of family groups across Great Britain, using historic censuses and recent consumer registers. The GeoDS holds a number of novel consumer datasets that chart recent movements and changes, including the links between geographical and social mobility.
«
Type in a surname, and it will show where that name has been found. Intriguing.
unique link to this extract
London drug dealers are hijacking Google Maps • London Centric
Jim Waterson, Polly Smythe, Holly Wyche, and Cormac Kehoe:
»
Type the word ‘pharmacy’ into Google Maps, and you could wind up with more than a place to pick up an NHS prescription. That’s because London’s drug dealers are registering themselves using the term, at unrelated addresses, in order to advertise their services.
We found dealers who claimed to sell everything from cocaine and ketamine to fake banknotes, cannabis, and guns. Finding a drug dealer’s number on the internet is hardly surprising. But the overlap with real-world London addresses is causing confusion for both would-be purchasers and legitimate property owners.
One London Centric reader who got in touch to flag the issue was trying to find a late night pharmacy that was open so she could buy deworming medicine for her child, only to find various offers of MDMA delivery. She said she was tempted by the alternative on offer but it might not have had the required effect.
While some dealers appeared to be using the physical pin on a map as a way to advertise their number or web address to people in that area, others gave the impression of being real shops that could be visited. A Google review from one disappointed customer said they had travelled to a physical central London location building to buy drugs after finding it on Google Maps only to find an office building. Using what appears to be their real name, they gave the listing one star and declared “there’s no shop when you get there”.
Many of the listings used photos of fake shopfronts that are AI-generated, including one that said “Weed Shop” in colourful lettering. Another claimed to be a large two-storey black building with a gigantic marijuana leaf over its door on the Thames riverside between Westminster Bridge and Blackfriars Bridge. When we searched on Google Maps for a pharmacy in Soho we were offered the chance to buy a gun.
«
Never used to get that with the printed London A-Z. (London Centric, meanwhile, goes from strength to strength.)
unique link to this extract
Universities sound warnings about AI even as they embrace it • The New York Times
Mark Arsenault, Dana Goldstein,Alan Blinder and Sarah Mervosh:
»
For a motivated student, an AI chatbot can function as a sort of tutor. Chatbots can break down the steps in a math problem, surface unexpected sources for research papers and offer feedback on writing.
Still, a growing body of research suggests that AI use can also harm learning.
A recent study of 26,000 Chinese secondary-school students found that AI adoption led to an 18% increase in homework scores, but a 20% decline in scores on secure exams. The implication was that rushing through homework with help from chatbots degraded the deeper learning needed to perform well when the technology was not available.
ImageOne 2025 study out of MIT found that people who used chatbots as writing assistants struggled to remember what they had written only minutes after finishing their essay. “You basically didn’t engage any of the memory networks that would actually help you remember,” said Nataliya Kos’myna, a MIT researcher who worked on the study.
Other research has shown that people who have greater trust in AI and rely on it more are more likely to have lower cognitive engagement.
But AI can be hard to avoid. Chatbots are embedded within the Google Classroom suite, which is nearly ubiquitous in schools, and within many online learning systems used by universities.
Some tools block outright cheating or plagiarism and allow professors to toggle specific A.I. capabilities on and off. But the most popular chatbots among students will instantly complete entire assignments, including research papers, on a student’s behalf.
Sian Leah Beilock, the president of Dartmouth, has argued that preparing graduates to use A.I. is an existential requirement of institutions of higher education.
“In a world that promises to be defined by AI, universities that fail to produce graduates who can use it productively will consign themselves to irrelevance,” she wrote recently.
«
Oh, well done Google, how helpful of you. But you can see the bind that universities and schools are in.
unique link to this extract
The American age is over • The Atlantic
Phillips Payson O’Brien:
»
The American age that followed World War II was bound to end eventually. What’s shocking is how suddenly that moment has arrived. In the past six months, America’s power—long exemplified by the world’s best warfighting equipment and most highly trained military personnel, a global network of bases and logistics systems, and military and diplomatic partnerships with all corners of the world—has proved insufficient to defeat Iran, a repressive, economically stagnant regime armed with cheap missiles and drones.
For years, virtually everyone thought that, in a crisis, Washington could draw on deeper reserves of money, troops, munitions, and diplomatic support than any other nation. Even when US forces faced setbacks in conflicts such as Vietnam, Iraq, and Afghanistan, a widespread assumption about America’s fundamental strength reassured allies and deterred foes. Yet that assumption also promoted a complacency that has left the US military exposed now that even inexpensive, mass-produced weaponry can be guided with deadly accuracy.
President Trump did not erode US capabilities on his own. But Trump’s war with Iran has revealed just how the underlying dynamics of global power have shifted—and just how much American might has decayed.
…For years, the United States has focused on building so-called exquisite systems—advanced weaponry that is produced in limited quantities at a high per-unit cost. This approach was in part an adaptation to the post-9/11 War on Terror, in which US forces focused on counterinsurgency and fought enemies that lacked the capacity to mount and sustain serious attacks on American logistics. Under those circumstances, the Pentagon built up a broad array of munitions but not deep stockpiles of them. At current production rates, the United States will need several years to build supplies back to where they were when Trump went to war in February—and even those levels in hindsight look far too weak to maintain America’s long-standing global position amid a shift toward mass-produced precision-guided drones.
The erosion of American manufacturing superiority is an old story, but it is a key element for understanding the end of the American age. The US could once outbuild other powers by tapping into civilian production. During World War II, Americans retooled the automotive industry to build aircraft. Today, relative to the rest of the world, the United States does almost no civilian shipbuilding. China, in comparison, makes about 50% of the ships in the world and controls about 80% of the world’s drone manufacture. Having built its military to prevail in a burst of high-intensity fighting, the United States simply lacks the equipment, stockpiles, and rested forces to engage China for any extended period of time.
«
(Gift link.)
unique link to this extract
Gas prices are about to take a big jump, analysts say, with the worst still to come • The Washington Post
Evan Halper, Rachel Chason and Joyce Sohyun Lee:
»
The effects of the U.S.-led war on Iran are about to intensify in the form of huge jumps in the prices of gasoline and diesel, analysts predict. And despite the Trump administration’s assurances, a new and potentially even more economically damaging phase of the war may be about to begin.
The national average price for regular gasoline stood at $4.44 on Thursday, according to AAA, with a range of $6.09 in California to $3.92 in Indiana. That’s still below the all-time national record of $5.02 in June 2022. But diesel, which has a big influence on trucking and retail prices, has continued to hit records, with the national average at $6.40 on Thursday. In California, diesel is averaging $8.35.
The cost increases are landing particularly hard in several political battleground states in the Midwest, where prices jumped as much as 25 cents per gallon overnight, according to AAA.
Damage inflicted on Saudi Arabia’s East-West pipeline last week, along with continuing strikes on oil infrastructure and contested control over a key waterway, threaten to elevate what had been a shipping crisis into a broader disruption of oil.
The 745-mile pipeline is Saudi Arabia’s principal overland alternative to the Strait of Hormuz, linking eastern oil fields to Red Sea export terminals.
U.S. Energy Secretary Chris Wright said Tuesday that the pipeline would be operational again in a matter of days, and the White House has touted renewed tanker traffic through the Strait of Hormuz.
But analysts who reviewed images of the pipeline damage anticipate a significantly longer disruption, likely resulting in high prices for consumers well beyond the midterm elections.
«
The single most important metric in polling ahead of elections? The price of fuel.
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 