The demand for RAM for AI datacentres means less will be available for cheap smartphones – which means their sales will crash in 2026. CC-licensed photo by Asian Development Bank on Flickr.
A selection of 10 links for you. Remember? I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.
Instagram to alert parents if teens search for self-harm content • BBC News
Richard Morris and Liv McMahon:
»
Parents using Instagram’s child supervision tools will soon receive alerts if their teen repeatedly searches for suicide or self-harm related terms on the platform.
It is the first time parent company Meta will proactively alert parents to searches by their child on Instagram for harmful material, rather than just block searches and direct users to external help.
Parents and teens enrolled in Instagram’s Teen Accounts experience in the UK, US, Australia and Canada will be notified about the alerts from next week, with the rest of the world to follow later.
But suicide prevention charity the Molly Rose Foundation has strongly criticised the measures, warning they “could do more harm than good”.
“This clumsy announcement is fraught with risk and we are concerned that forced disclosures could do more harm than good,” said its chief executive Andy Burrows.
The organisation was established by the family of Molly Russell, who took her own life in 2017 at the age of 14 after viewing self-harm and suicide content on platforms including Instagram.
Burrows said “every parent would want to know if their child is struggling, but these flimsy notifications will leave parents panicked and ill-prepared to have the sensitive and difficult conversations that will follow”.
Meta says alerts to parents about their child searching for suicide and self-harm material within a short space of time on Instagram will also be accompanied by expert resources to help them navigate difficult conversations.
However, Molly Russell’s father Ian, who set up the Molly Rose Foundation in her honour, remains sceptical about the alerts. “Imagine being a parent of a teenager and getting a message at work saying ‘your child is thinking of ending their life’… I don’t know how I’d react,” he told the BBC.
…A number of charities including the Molly Rose Foundation have said Meta’s announcement is almost an acknowledgment that more could be done to protect children on Instagram.
«
It’s incredibly slow movement by Instagram. And the responsibility for removing dangerous content remains with.. Instagram.
unique link to this extract
Claude Blattman · AI for Researchers & Managers • Chris Blattman
Chris Blattman is a political economist at UChicago Harris:
»
In January 2026 I started building AI workflows with Claude Code. What seemed impossible then — inbox triage, meeting capture, proposal drafting, project dashboards, trip planning — is running today. “Claude Blattman” isn’t real, but the tools are. I’ve never coded in my life, so if I can do this you can.
…Like most academics, the work that matters — the research, the writing, the thinking — gets buried under an avalanche of email, coordination, proposals, and administrative overhead. I manage a large portfolio of research projects simultaneously, each with distributed teams across multiple countries. I spend more time answering emails and writing grant reports than doing the science I was trained for.
For the past year I intensively used AI chatbots. They were invaluable — deep research, better drafts, faster brainstorming, decent writing feedback. But they were limited. The time savings were real but modest.
Then in January 2026 I discovered Claude Code, and the center of gravity shifted from chatbots to agentic tools — AI that can read your files, search your email, manage your calendar, and build workflows that improve themselves over time.
I spent hours daily for a month building skills and processes. The time investment hasn’t paid off yet in pure hours saved, but I can see where this is going — and things that seemed impossible are working today.
…I built this site because I’m in the same position as most of my readers: buried in admin, not a developer, trying to claw back time for work that actually matters.
If a non-technical professor can build a working AI workflow system, you can too.
«
This is a useful primer with real working examples. The email triage is obviously useful; I would have loved to have had that when I was getting 200+ emails per day as a daily journalist. If you’re feeling like you’re drowning in unimportant stuff, this is surely worth reading.
unique link to this extract
Burger King will use AI to check if employees say ‘please’ and ‘thank you’ • The Verge
Emma Roth:
»
Burger King is launching an AI chatbot that will live in the headsets used by employees. The voice-enabled chatbot, called “Patty,” is part of an overarching BK Assistant platform that will not only assist employees with meal preparation but also evaluate their interactions with customers for “friendliness.”
Thibault Roux, Burger King’s chief digital officer, tells The Verge that the company compiled information from franchisees and guests on how to measure friendliness, resulting in the fast food chain training its AI system to recognize certain words and phrases, such as “welcome to Burger King,” “please,” and “thank you.” Managers can then ask the AI assistant how their location is performing on friendliness. “This is all meant to be a coaching tool,” Roux says, adding that the company is “iterating” on capturing the tone of conversations as well.
The OpenAI-powered Patty serves as the “voice” of the BK Assistant platform, which combines data across drive-thru conversations, kitchen equipment, inventory, and other areas of the Burger King business. Employees can ask Patty questions, such as how many strips of bacon to put on a Maple Bourbon BBQ Whopper, or for instructions on how to clean the shake machine.
Because it’s integrated with the new cloud point-of-sale system, the AI assistant will also alert managers if a machine is down for maintenance or when an item is out of stock. “Within 15 minutes, the entire ecosystem will remove it from stock — whether you’re walking into a restaurant to order from the kiosk, whether you’re going to the drive-thru, the digital menu board will be updated,” Roux says.
«
Thanks, I hate it. A micromanager that isn’t even human.
unique link to this extract
Worldwide smartphone market to decline 13% in 2026, its largest ever drop • IDC
»
Worldwide smartphone shipments are forecast to decline 12.9% year-on-year (YoY) in 2026 to 1.12bn units, according to the International Data Corporation (IDC) Worldwide Quarterly Mobile Phone Tracker. This decline will bring the smartphone market to its lowest annual shipment volume in more than a decade. The current forecast represents a sharp decline from our November forecast amid the intensifying memory shortage crisis.
“What we are witnessing is not a temporary squeeze, but a tsunami-like shock originating in the memory supply chain, with ripple effects spreading across the entire consumer electronics industry,” said Francisco Jeronimo, vice president for Worldwide Client Devices, IDC. “The global smartphone market, particularly Android manufacturers, faces a significant threat. Vendors whose business is mainly at the low end of the market are likely to suffer the most. Rising component costs will hit their margins, and they will have no choice but to pass the costs on to end users.
“By contrast, Apple and Samsung are better positioned to navigate this crisis. As smaller and low-end-positioned Android vendors struggle with rising costs, Apple and Samsung could not only weather the storm but potentially expand market share as the competitive landscape tightens.”
“The memory crisis will cause more than a temporary decline; it marks a structural reset of the entire market, fundamentally reshaping long‑term TAM (Total Addressable Market), the vendor landscape, and the product mix,” said Nabila Popal, senior research director with IDC’s Worldwide Quarterly Mobile Phone Tracker.
“We expect consolidation as smaller players exit, and low-end vendors face sharp shipment declines amid supply constraints and lower demand at higher price points. Although shipments will witness a record drop, Smartphone ASP [average selling price] is projected to rise 14% to a record $523 this year.
«
This is a lot of people putting off buying a new phone, especially at the low end, where a panoply of Android manufacturers are getting by in China and India by offering low-end products. Now their RAM supply is going to dry up, and so are they. The AI blast radius gets bigger.
unique link to this extract
The death of Spotify: why streaming is minutes away from being obsolete • The Artist Economy
Joel Gouveia:
»
I’m going to take the diplomatic hat off here and say with brutal honesty: basically everybody in the music business hates Spotify except for the people who work there. It’s a platform that sucks artists for everything they have, it actively prevents community building, and, despite all of that, the platform still struggles to maintain a healthy profit margin.
The streaming business model is fundamentally broken. And eventually, its demise will become more and more obvious to recognize. I’ll break down exactly why the DSP era is coming to a grinding halt, why the major labels are quietly terrified, and why the artists who don’t pivot now are going to go down with the ship.
…[Jimmy] Iovine put it bluntly: “The streaming services have a bad situation, there’s no margins, they’re not making any money.”
This model only works for Apple, Amazon, and Google, because they don’t need their music platforms to be wildly profitable. Amazon uses music as a loss-leader to keep you paying for Prime. Apple uses it to sell $1,000 iPhones.
As for Spotify, or any standalone music streaming company, they’re kind of screwed. And guess what – when the platform’s margins are structurally squeezed, guess who gets squeezed first? The artists.
…If the DSPs are “minutes away from obsolete,” what replaces them? Well, I’m not sure the DSPs are going to disappear overnight, but if you’re an artist or a manager trying to sustain yourself in this evolving music economy, the answer is direct ownership.
The artists who will survive the next five years are the ones who are quietly shifting their focus away from the “ATM Machine.” They are building their own cultural hangars.
They are capturing phone numbers on Laylo. They are driving fans to private Discord servers. They are focusing on ARPF (Average Revenue Per Fan) through high-margin merch, vinyl, and hard tickets, rather than begging for fractions of a penny from a playlist placement.
We are witnessing the death of the “Mass Audience” and the birth of the “Micro-Community.”
«
Large-scale online deanonymization with LLMs • ArXiv
Simon Lermen, Daniel Paleka et al:
»
We show that large language models can be used to perform at-scale deanonymization. With full Internet access, our agent can re-identify Hacker News users and Anthropic Interviewer participants at high precision, given pseudonymous online profiles and conversations alone, matching what would take hours for a dedicated human investigator. We then design attacks for the closed-world setting.
Given two databases of pseudonymous individuals, each containing unstructured text written by or about that individual, we implement a scalable attack pipeline that uses LLMs to: (1) extract identity-relevant features, (2) search for candidate matches via semantic embeddings, and (3) reason over top candidates to verify matches and reduce false positives. Compared to prior deanonymization work (e.g., on the Netflix prize) that required structured data or manual feature engineering, our approach works directly on raw user content across arbitrary platforms.
We construct three datasets with known ground-truth data to evaluate our attacks. The first links Hacker News to LinkedIn profiles, using cross-platform references that appear in the profiles. Our second dataset matches users across Reddit movie discussion communities; and the third splits a single user’s Reddit history in time to create two pseudonymous profiles to be matched.
In each setting, LLM-based methods substantially outperform classical baselines, achieving up to 68% recall at 90% precision compared to near 0% for the best non-LLM method. Our results show that the practical obscurity protecting pseudonymous users online no longer holds and that threat models for online privacy need to be reconsidered.
«
One of the authors is from Anthropic. The response of Matthew Green, the cryptographer at Johns Hopkins University, was that this wrecks pseudonymity online. It might rely more on written text than content posted to TikTok or Instagram, but those might be the people you’re trying to deanonymise.
unique link to this extract
AI mistakes are infuriating gamers as developers seek savings • Bloomberg
Vlad Savov:
»
A backlash against the spreading use of AI in video games has grown “sensational,” the developer behind the surprise smash hit Arc Raiders has warned after being caught in the middle of the fallout.
The $200bn video game industry is riven by disagreement over how to integrate AI into its creative processes. The new tech has been celebrated as the next big revolution by some, but also lambasted as a threat driving out human creativity and degrading quality. The very people who might have been expected to be its most avid fans, PC gamers, are obstinately hostile.
Arc Raiders sold 12m copies in three months after its release last year and is the most-played paid game on PC platform Steam today. But, for a brief time, it was vilified online by players for including robotic-sounding auto-generated voices.
Patrick Soderlund, who set up Stockholm-based Embark Studios AB after steering major franchises as a senior executive at Electronic Arts, said the Arc Raiders team leaned on AI for the parts that weren’t essential for immersive play. They used a mix of professional actors and automated voices.
“We started with a thesis and the foundation that the industry is troubled,” Soderlund, 52, said of the company’s origins. “We saw escalating development costs that made it very difficult to make games affordable.”
Burdened by the sins of recent past — overhiring during the pandemic boom and letting production costs spiral out of control — game developers have been shedding jobs and canceling projects. Sony Group Corp. last week shut down Austin-based Bluepoint Games, which it acquired five years earlier. Paris-based Ubisoft Entertainment SA in January announced a complete reorganization that sank its share price by 40% in a day, though it expressed hope that generative AI could be a salve.
…That message has struggled to reach gamers, who’ve built up a cynicism about big game publishers after years of flops, mismanagement and layoffs at fan-favorite studios. Part of the problem stems from the immaturity of AI tools, whose product is only ever apparent to players when it’s worse than human work — including the voices in Arc Raiders.
The fans’ worry about AI squeezing the art out of their favorite pastime is increasingly shared by industry insiders. Some 47% of developers polled by research house Omdia last year said they expect generative AI use to reduce game quality, with views souring from a year earlier.
«
(The link is free for a week, apparently. Get it while it’s hot!)
unique link to this extract
New York sues video game developer Valve, says its ‘loot boxes’ are gambling • Reuters
Jonathan Stempel:
»
New York’s attorney general sued Valve, a video game developer whose franchises include Counter-Strike, Team Fortress and Dota, accusing it of promoting illegal gambling and threatening to addict children through its use of “loot boxes.”
In a complaint filed on Wednesday in a state court in Manhattan, Attorney General Letitia James said Valve’s loot boxes amounted to “quintessential gambling,” violating the state’s constitution and penal law, with valuable items often hard to win and many items worth pennies. Valve, based in Bellevue, Washington, did not immediately respond to requests for comment.
Loot boxes let players use real money to buy chances to win virtual items, such as decorations for characters and weapons, in an effort to convey status.
James said Valve generated billions of dollars of revenue by selling “keys” to open loot boxes, including in one game where the process resembled a slot machine as a wheel whirred through various items before stopping. The attorney general said key sales advanced Valve’s unusual business model of letting players sell items they won on its virtual marketplace, Steam Community Market, and on other marketplaces.
“Valve’s loot boxes are particularly pernicious because they are popular among children and adolescents,” according to the complaint.
Children introduced to gambling by age 12 are four times more likely to become problem gamblers as adults, the complaint added, citing the Massachusetts Department of Public Health. James is seeking restitution for players, plus a fine of three times Valve’s alleged illegal gains.
Loot boxes for video games have been the subject of other regulatory action. For example, in January 2025, the US Federal Trade Commission fined, opens new tab Singapore-based Cognosphere, the maker of Genshin Impact, $20m for deceiving children and others about the odds of winning valuable loot-box prizes.
«
Seems like an open-and-shut case. Do they encourage people to spend money on a random outcome? Yes. The end.
unique link to this extract
‘An AlphaFold 4’ – scientists marvel at DeepMind drug spinoff’s exclusive new AI • Nature
Ewen Callaway:
»
Nearly two years after Google DeepMind released an updated AlphaFold3 geared at drug discovery, its biopharmaceuticals spin-off, Isomorphic Labs, announced an even more powerful artificial-intelligence model — and they’re keeping it all to themselves.
Isomorphic Labs, based in London, touted the capacities of its “drug-discovery engine” — which it calls IsoDDE — in a 27-page technical report, released on 10 February. Achievements, including precise predictions of how proteins interact with potential drugs and antibody structures, have impressed scientists working in the field.
Yet unlike the AlphaFold AI systems for predicting protein structure — which were made accessible to other researchers and described in depth in journal articles — IsoDDE is proprietary, and the technical paper offers scant insight into how to achieve similar results.
“It’s a major advance, on the scale of an AlphaFold4,” referring to an unreleased future generation of Google DeepMind’s technology, says Mohammed AlQuraishi, a computational biologist at Columbia University in New York City who is working to develop fully open-source versions of AlphaFold. “The problem, of course, is that we know nothing of the details.”
«
The full article requires a (free) login. AI is starting to push further into biotech: a couple I was talking to recently who work in biotech said that AlphaFold is the biggest thing that has happened in the space in the past few years.
unique link to this extract
Exclusive: “rare earth” element shortages worsen in US aerospace, chips despite trade truce, sources say • Reuters
Allison Lampert, Laurie Chen, Lewis Jackson and Michael Martina:
»
Suppliers to US aerospace and semiconductor firms face worsening rare earth shortages, with two turning away some clients, industry insiders said, weeks before US President Donald Trump is expected to meet his Chinese counterpart Xi Jinping for a summit in Beijing.
The shortages center on rare earths such as yttrium and scandium, niche members of the family of 17 elements, which play tiny but vital roles in defence technology, aerospace and semiconductors and are almost entirely produced in China.
While Beijing has allowed many rare earth exports to resume since it imposed restrictions in April, shipments of these materials still rarely make it to the U.S. despite the October detente with Washington, Chinese customs data show.
That easing of trade tensions, premised in part on China pausing its critical mineral export restrictions, will be on the table when Trump and Xi meet in Beijing in March.
A key pain point is yttrium, used in coatings that keep engines and turbines from melting at high temperatures. Without regular application of these coatings, engines cannot be used.
Since Reuters first reported about yttrium shortages in November, prices have jumped 60% and are now about 69 times as high as a year ago. Some coatings manufacturers are also now starting to ration material, according to company executives and traders.
…Another firm in the coating supply chain recently ran out of material and stopped selling products containing yttrium oxide, according to a source with direct knowledge of the matter.
While shortages of yttrium and scandium have not weighed on production of jet engines or chips yet, a US government official told Reuters some US manufacturers now face “shortages” of certain rare earths from China.
«
| • 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
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.
The Overspill: when there's more that I want to say 