In January 2022, a volcano near the remote island of Tonga erupted – and cut the island off from the internet, and the rest of the world. CC-licensed photo by James St. John 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.

---

There’s another post coming this week at the Social Warming Substack on Friday at 0845 UK time. Free signup.

---

A selection of 10 links for you. Did it? I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.

---

UK once again demands backdoor to Apple’s encrypted cloud storage • Financial Times via Ars Technica

Anna Gross and Tim Bradshaw:

»

The UK government has issued a new order to Apple to create a backdoor into its cloud storage service, this time targeting only British users’ data, despite US claims that Britain had abandoned all attempts to break the tech giant’s encryption.

The UK Home Office demanded in early September that Apple create a means to allow officials access to encrypted cloud backups, but stipulated that the order applied only to British citizens’ data, according to people briefed on the matter.

A previous technical capability notice (TCN) issued in January sought global access to encrypted user data. That move sparked a diplomatic clash between the UK and US governments and threatened to derail the two nations’ efforts to secure a trade agreement.

In February, Apple withdrew its most secure cloud storage service, iCloud Advanced Data Protection, from the UK.

“Apple is still unable to offer Advanced Data Protection in the United Kingdom to new users,” Apple said on Wednesday. “We are gravely disappointed that the protections provided by ADP are not available to our customers in the UK given the continuing rise of data breaches and other threats to customer privacy.”

It added: “As we have said many times before, we have never built a back door or master key to any of our products or services and we never will.”

The Home Office said: “We do not comment on operational matters, including, for example, confirming or denying the existence of any such notices.

“We will always take all actions necessary at the domestic level to keep UK citizens safe.”

Both Apple and the Home Office are restricted from discussing TCNs by law.

«

As the eminent cryptographer Matthew Green points out, this is a really strange demand. ADP (user-encrypted backups) aren’t available in the UK. Only health data, web history and passwords are end-to-end encrypted in backups. Or iMessage, maybe. Or, final maybe, the ADP backups of foreign users on British soil. (Spies? Terrorists?)

It’s all terrible overreach which makes the government look bad. Or worse.
unique link to this extract

---

Intel in early talks to add AMD as foundry customer • Semafor

Rohan Goswami:

»

Intel is in early-stage talks to add AMD as a customer at Intel’s factories, in what would be another vote of confidence in the struggling chipmaker, according to people familiar with the matter.

In the past seven weeks, Intel has gained investment dollars and public support from the White House, Nvidia, and SoftBank, and is in talks for backing from Apple, Semafor and others have reported. AMD designs chips that are currently produced mostly by Taiwan’s TSMC, and Intel currently lacks the technology to produce AMD’s most advanced, profitable chips.

It’s unclear how much of their manufacturing would shift to Intel if the two companies reach a deal, or whether it would come with a direct investment by AMD, similar to the deals cut by other companies. It is possible that no agreement will be reached, the people said.

Spokespeople for Intel and AMD declined to comment. Intel shares rose around 3.5% on the news and are up around 77% for the year.

AMD has reason to stay in the White House’s good graces. Its significant business selling chips in China was hit by export restrictions earlier this year, which Trump recently loosened.

Intel’s chip factories are considered inferior to TSMCs. But big American companies, following the Trump administration’s preference for having a US chip-manufacturing champion, have diverted at least some of their production, mostly for less-advanced chips, towards Intel’s domestic foundries.

«

Not impossible, only unlikely. And becoming more likely all the time.
unique link to this extract

---

OpenAI’s Sora 2 lets users insert themselves into AI videos with sound • Ars Technica

Benj Edwards:

»

Aside from visual and auditory upgrades, OpenAI is taking another big step away from its AI research lab pedigree toward making the new model available to average people in an easy-to-use way. It’s doing it by packaging Sora 2 into a social iOS app that focuses on creating and sharing AI-generated content.

That new iOS app has already launched in the US and Canada as an invite-based rollout, with plans to expand to additional countries. Users can sign up in the app for notifications when access becomes available for their account. The service will initially be free with what OpenAI describes as “generous limits,” though the company plans to offer paid options for additional generations when demand exceeds available compute resources.

Using the app, users can create videos, remix content from other users, and browse a customizable feed of generated videos. As mentioned above, the app’s Cameo feature allows users to essentially deepfake themselves by recording a one-time video and audio capture, which the model can then insert into any Sora-generated scene.

In addition to the basic Sora 2 model on the website and in the app, ChatGPT Pro subscribers will gain access to Sora 2 Pro, described as an experimental higher-quality model. OpenAI also plans to release Sora 2 through its API for developers. The older Sora 1 Turbo model will remain available, and existing creations will stay in users’ Sora libraries.

So, what could go wrong with an app that can easily put people into AI-generated videos? Well, just about everything. Battling misuse is likely going to be a tricky issue for the company. In the recent past, we’ve seen instances of AI deepfaking (not related to OpenAI) without consent that have led to bullying lawsuits, criminal penalties, and suicides.

OpenAI is taking precautions. Given recently prominent corporate sensitivities following a ChatGPT user’s suicide, OpenAI says Sora 2 includes specific protections for teenage users. Those include default daily-generation limits and strict permissions for the cameos feature. OpenAI says it has deployed both automated safety systems and human moderators to review potential cases of bullying or misuse.

In particular, OpenAI has built in layers of security for the cameos feature. It says that users can maintain control over their uploaded likeness: They can decide who can use their cameo in videos and can revoke access or remove videos containing their likeness at any time. Users can also view all videos containing their cameo, including drafts created by other people.

«

OpenAI is talking nonsense here. There are videos including OpenAI CEO’s Sam Altman’s likeness all over the place.
unique link to this extract

---

People are farming and selling Sora 2 invite codes on eBay • 404 Media

Jason Koebler:

»

People are farming and selling invite codes for Sora 2 on eBay, which is currently the fastest and most reliable way to get onto OpenAI’s new video generation and TikTok-clone-but-make-it-AI-slop app. Because of the way Sora is set up, it is possible to buy one code, register an account, then get more codes with the new account and repeat the process.

On eBay, there are about 20 active listings for Sora 2 invite codes and 30 completed listings in which invite codes have sold. I bought a code from a seller for $12, and received a working code a few minutes later. The moment I activated my account, I was given four new codes for Sora 2. When I went into the histories of some of the sellers, many of them had sold a handful of codes previously, suggesting they were able to get their hands on more than four invites. It’s possible to do this just by cycling through accounts; each invite code is good for four invites, so it is possible to use one invite code for a new account for yourself, sell three of them, and repeat the process.

There are also dozens of people claiming to be selling or giving away codes on Reddit and X; some are asking for money via Cash App or Venmo, while others are asking for crypto. One guy has even created a website in which he has generated all 2.1 billion six-digit hexadecimal combinations to allow people to randomly guess / brute force the app (the site is a joke). 

«

I get the feeling OpenAI isn’t that worried about more people using Sora 2. But also: where’s the evidence that OpenAI is working towards its fabled AGI, rather than a video slop machine?
unique link to this extract

---

Critics slam OpenAI’s parental controls while users rage, “treat us like adults” • Ars Technica

Ashley Belanger:

»

As OpenAI tells it, the company has been consistently rolling out safety updates ever since Matthew and Maria Raine, sued OpenAI, alleging that ChatGPT led to the death of their son.

On August 26, the day that the lawsuit was filed, OpenAI seemed to publicly respond to claims that ChatGPT acted as a “suicide coach” for 16-year-old Adam Raine by posting a blog promising to do better to help people “when they need it most.”

By September 2, that meant routing all users’ sensitive conversations to a reasoning model with stricter safeguards, sparking backlash from users who feel like ChatGPT is handling their prompts with kid gloves. Two weeks later, OpenAI announced it would start predicting users’ ages to improve safety more broadly. Then, this week, OpenAI introduced parental controls for ChatGPT and its video generator Sora 2. Those controls allow parents to limit their teens’ use and even get access to information about chat logs in “rare cases” where OpenAI’s “system and trained reviewers detect possible signs of serious safety risk.”

While dozens of suicide prevention experts in an open letter credited OpenAI for making some progress toward improving safety for users, they also joined critics in urging OpenAI to take their efforts even further, and much faster, to protect vulnerable ChatGPT users.

Jay Edelson, the lead attorney for the Raine family, told Ars that some of the changes OpenAI has made are helpful. But they all come “far too late.” According to Edelson, OpenAI’s messaging on safety updates is also “trying to change the facts.”

…On the X post where OpenAI announced parental controls, some parents slammed the update.

In the X thread, one self-described parent of a 12-year-old suggested OpenAI was only offering “essentially just a set of useless settings,” requesting that the company consider allowing parents to review topics teens discuss as one way to preserve privacy while protecting kids.

«

The tension between child (or teenager) safety and adult use will always exist where there’s a single interface to the whole product. You simply can’t pretend they won’t intersect. But OpenAI (like YouTube before it) is pretending that.
unique link to this extract

---

Internet and cell phone resume in Afghanistan​ • Reuters

Mohammad Yunus Yawar:

»

Cell phone and internet services were restored in Afghanistan on Wednesday, local residents said, some 48 hours after diplomatic and industry sources said connectivity was abruptly cut on the orders of the Taliban administration.

The cell phone services of Roshan and Etisalat companies, the foreign-owned biggest providers, came back to life in the late afternoon, residents in Kabul and other cities said. Internet access was restored, according to companies providing the service.

A Taliban official from the information department said there were technical reasons for the outage and that services would be quickly restored. He did not immediately respond to a request for comment on whether the Taliban had ordered the outage.

…In the past, the Taliban have voiced concern about online pornography and authorities have cut fibre-optic links to some provinces in recent weeks, with officials citing morality concerns.

The outage on connectivity, which started on Monday, follows a series of hardline strictures this year, as the Taliban’s conservative leadership, based in the southern city of Kandahar, enforces its views in a tussle against some relatively more open-minded ministers in the capital Kabul.

The outage had caused chaos, with financial remittances, trade with neighbouring countries and the operations of banks paralysed, while many Afghans were left stranded without flights.

Online learning by teenage girls and women, an education lifeline after they were banned by the Taliban from high schools and universities, was also brought to a stop.

«

These “morality concerns” – wouldn’t the simplest thing be to ban men from using the internet, and insist that only women could? Can the Taliban go that far?
unique link to this extract

---

Please let the robots have this one • The Argument Mag

Kelsey Piper:

»

In The Argument’s September polling on all things AI, we asked “Do you think your city or town should allow or ban self-driving cars?” Twenty-eight% wanted to allow them. Forty-one% wanted to ban them. There were no truly enthusiastic contingencies for self-driving cars — they do a bit better with liberals, younger people, and more educated people, but not a single group has a majority in favor.

By contrast, there’s a large mass of old people that are extremely opposed to self-driving cars, which is especially ironic when you consider that they are also the most likely to cause the very crashes that this technology would protect us from.

When we asked voters why they want to ban self-driving cars, they gave us a range of reasons, which generally boiled down to safety and trust.

Hopefully, the safety concerns will be responsive to the safety data! To most of the country, self-driving cars are still a hypothetical, and I think at least some people will be won over when they see how well these cars drive (as a pedestrian, it’s really notable how they just stop for you at crosswalks!). And there is more support for self-driving cars out West, where they actually exist in significant numbers.

But even in San Francisco where Waymos are a reality, I still run into a lot of Waymo haters. Sometimes they hate the cars for taking jobs from Uber and Lyft drivers. (Though remember how a couple of years ago everyone wanted to ban Uber and Lyft?) Sometimes they hate the cars out of vague anti-corporate sentiment. But often, what they really hate isn’t Waymo at all.

A Waymo is often seen as a physical-world embodiment of the flood of AI slop replacing high-quality artwork and filling social media feeds with fake nonsense. That quiet, whirring car is a reminder that the once-useful assumption that a lengthy written text had required some human to apply serious thought at some point is now no longer valid. But you can’t vandalize ChatGPT, so anti-AI sentiment finds its expression in harassing Waymos.

«

There is the little fact about self-driving cars not really being a thing in more complicated road conditions, but for American cities they make a lot of sense.
unique link to this extract

---

‘Reverse Midas touch’: Starmer plan prompts collapse in support for digital IDs • The Guardian

Eleni Courea:

»

Public support for digital IDs has collapsed after Keir Starmer announced plans for their introduction, in what has been described as a symptom of the prime minister’s “reverse Midas touch”.

Net support for digital ID cards fell from 35% in the early summer to -14% at the weekend after Starmer’s announcement, according to polling by More in Common. The findings suggest that the proposal has suffered considerably from its association with an unpopular government. In June, 53% of voters surveyed said they were in favour of digital ID cards for all Britons, while 19% were opposed.

Starmer set out plans to roll out a national digital ID scheme on Friday, saying it presented an “enormous opportunity” for the UK that would “make it tougher to work illegally in this country”.

Just 31% of people surveyed after Starmer’s announcement over the weekend said they were supportive of the scheme, with 45% saying they were opposed. Of those, 32% said they were strongly opposed. More than 2.6 million people have signed a petition against introduction of the IDs.

Advocates of a national digital ID scheme are frustrated at the way the policy has been presented and believe that now it may never be implemented.

More in Common’s polling suggests public dissatisfaction with the government is behind the collapse in support. It found that 58% of those who thought Starmer was doing a bad job as prime minister opposed digital IDs, while only 20% supported them. On the other hand, of those who thought Starmer was doing a good job, 71% were supportive of digital IDs and 14% were opposed.

…Ministers have said digital ID cards will be used to prove a person’s right to live and work in the UK and will be compulsory for anyone who wants employment. The government is concerned that the relative ease of finding illegal work in the UK’s shadow economy is one of the factors encouraging people to make illegal and dangerous journeys across the Channel.

The photo IDs will be stored on smartphones in a similar way to digital bank cards and will contain information on the holder’s name, residency status, date of birth and nationality. They will not be required to access healthcare or welfare payments.

«

The idea is a bit barmy, though. Completely uncooked.
unique link to this extract

---

The latest on the Anthropic settlement with authors • Authors’ Licensing and Collecting Society (ALCS)

»

In June, the judge presiding over the Bartz v Anthropic lawsuit ruled that Anthropic must pay damages for its use of pirated datasets LibGen and the Pirate Library Mirror (PiLiMi) to train its AI models. 

A trial was scheduled for December to determine the extent of damages paid, which Anthropic has avoided by reaching a landmark settlement with the authors involved in the class-action lawsuit. This settlement has now been given preliminary approval. 

What happens next? By Thursday October 2, the website www.anthropiccopyrightsettlement.com will be made live. This website will include a searchable list of all works eligible for compensation. It will also include a claims form for those with eligible works. This is how you identify yourself and officially request payment. The website will also include a calendar of relevant deadlines.  

If your works are included in this list, you will receive an official notice from the Settlement Administrator. However, you don’t need to wait for this notice to submit your claims form. The deadline to claim is 23 March 2026. 

Authors based outside of the US are also eligible for payment. However, to be eligible for payment their works must: 

• Have been downloaded from LibGen and/or PiLiMi
• Have been registered with the US Copyright Office within 3 months of publication or before it was downloaded by Anthropic and within 5 years of publication 
• Have a valid ISBN or ASIN 

Because of the strict eligibility criteria, only approximately 500,000 titles out of the seven million copies of books that Anthropic reportedly downloaded meet the definition required to be included in the works list. Rightholders will receive approximately $3,000 per eligible title, with an optional split of 50/50 between author and publisher for most books, and a tailored approach for education works.

…The Authors Guild, who are administering this lawsuit, have compiled a comprehensive set of questions and answers on their website here.

«

Can’t decide if I want my works to have been pirated or not. But I’ll definitely go and check the website.
unique link to this extract

---

Extremely offline: what happened when a Pacific island was cut off from the internet • The Guardian

Samanth Subramanian:

»

Since its several brief eruptions the previous month, December 2021, Hunga Tonga-Hunga Ha‘apai had continued to gurgle and churn. On that Saturday, 15 January [2022], 2.4 cubic miles of sediment and molten rock shot through its mouth with the force of what scientists call a “magma hammer”, sending a plume of ash at least 35 miles up into the atmosphere. It was the largest atmospheric explosion that modern instruments had recorded, outdoing any nuclear bomb ever detonated. They heard the sound in Alaska. Seven and a half thousand miles away, in the south Indian city of Chennai, meteorologists measured an abrupt spike in atmospheric pressure. It was Hunga Tonga-Hunga Ha‘apai, doing its thing.

On his drive home, Vea had called relatives in the US through Facebook Messenger to let them know he was all right. At some point during their conversation, the line cut out. He assumed the network was overloaded by everybody getting online at the same time. “This is usually a problem for us,” he told me. Vea, DHL’s agent in Tonga, is the president of the Tonga Chamber of Commerce & Industry, and we met in his spare, sunny office in the capital of Nuku’alofa, three streets from the Pacific. The curtains were red, and the sun filtered through in a dull watermelon light.

Vea wears a perpetual expression of mirth, and it was difficult to imagine him as worried as he was on the day the volcano blew up, sitting in his van in the middle of a rain of ash, staring at his suddenly useless phone. He decided he’d try his relatives later, after the traffic online subsided. At home, though, the power was out, and he couldn’t charge his phone, so it was only the next morning, when he tuned in to Radio Tonga, that he learned the country had lost its internet altogether – and with it, all its means of reaching the world beyond the wide, silent water.

…Commerce broke down. Since this happened in the middle of the Covid pandemic, DHL was flying only one plane a week to Tonga – but without the internet, Vea couldn’t file or receive manifests online. The ATMs went dead, because banks couldn’t check how much money their customers had in their accounts – and that, in an economy still accustomed to cash, immediately put livelihoods in danger. Owners of fisheries and farms of squash and breadfruit were unable to fill out the compliance and quarantine forms needed to export their produce. Tongans living overseas couldn’t wire funds home to help their families – and at the time, foreign remittances made up 44% of the country’s GDP.

«

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