Start Up No.2347: OpenAI blames ‘glitch’ for Unsayable names, FTC blocks location data seller, a Verge subscription?, and more


The interface panels on Lego bricks can tell us a lot about good user interface design. CC-licensed photo by Scarlet Sappho 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 9 links for you. Plugged in. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.


ChatGPT’s refusal to acknowledge ‘David Mayer’ down to glitch, says OpenAI • The Guardian

Dan Milmo:

»

ChatGPT’s developer, OpenAI, has provided some clarity on the situation by stating that the Mayer issue was due to a system glitch. “One of our tools mistakenly flagged this name and prevented it from appearing in responses, which it shouldn’t have. We’re working on a fix,” said an OpenAI spokesperson

Some of those speculating on social media guessed the man at the centre of the issue was David Mayer de Rothschild, but he told the Guardian it was nothing to do with him and referenced the conspiracy theorising that can cluster around his family’s name online.

“No I haven’t asked my name to be removed. I have never had any contact with Chat GPT. Sadly it all is being driven by conspiracy theories,” he told the Guardian.

It is also understood the glitch was unrelated to the late academic Prof David Mayer, who appeared to have been placed on a US security list because his name matched the alias of a Chechen militant, Akhmed Chatayev.

However, the answer might lie closer to the GDPR privacy rules in the UK and EU. OpenAI’s Europe privacy policy makes clear that users can delete their personal data from its products, in a process also known as the “right to be forgotten”, where someone removes personal information from the internet. OpenAI declined to comment on whether the “Mayer” glitch was related to a right to be forgotten procedure.

OpenAI has fixed the “David Mayer” issue and is now responding to queries using that name, although other names that appeared on social media over the weekend are still triggering a “something appears to have gone wrong” response when typed into ChatGPT.

Helena Brown, a partner and data protection specialist at law firm Addleshaw Goddard, said “right to be forgotten” requests would apply to any entity or person processing that person’s data – from the AI tool itself to any organisation using that tool.

“It’s interesting to see in the context of the David Mayer issue that an entire name can be removed from the whole AI tool,” she said.

«

Certainly a lot of those suing OpenAI are going to have sat up and taken notice of this.
unique link to this extract


Here we go: The Verge now has a subscription • The Verge

Nilay Patel:

»

Okay, we’re doing this. 

Today [Tuesday] we’re launching a Verge subscription that lets you get rid of a bunch of ads, gets you unlimited access to our top-notch reporting and analysis across the site and our killer premium newsletters, and generally lets you support independent tech journalism in a world of sponsored influencer content. It’ll cost $7 / month or $50 / year — and for a limited time, if you sign up for the annual plan, we’ll send you an absolutely stunning print edition of our CONTENT GOBLINS series, with very fun new photography and design. (Our art team is delightfully good at print; we’ve even won a major magazine award for it.)

A surprising number of you have asked us to launch something like this, and we’re happy to deliver. If you don’t want to pay, rest assured that big chunks of The Verge will remain free — we’re thinking about subscriptions a lot differently than everyone else.

…we didn’t want to simply paywall the entire site — it’s a tragedy that traditional journalism is retreating behind paywalls while nonsense spreads across platforms for free. We also think our big, popular homepage is a resource worth investing in. So we’re rethinking The Verge in a freemium model: our homepage, core news posts, Decoder interview transcripts, Quick Posts, Storystreams, and live blogs will remain free. We know so many of you depend on us to curate the news every day, and we’re going to stay focused on making a great homepage that’s worth checking out regularly, whether you pay us or not.

Our original reporting, reviews, and features will be behind a dynamic metered paywall — many of you will never hit the paywall, but if you read us a lot, we’ll ask you to pay.

«

Interesting stats: ~500,000 people visit the site at least once a week, they read an average 14 stories per month (I make that one every two days), 55,000 come to the site every day.
unique link to this extract


We need to talk about dying • New Statesman

Rachel Clarke is a former journalist who re-trained to become a doctor and now specialises in palliative care; amid the debate about “assisted dying” she wants people to understand the reality, not the exaggeration, of death:

»

So called “ordinary dying” – a term deployed for decades by medical professionals to describe the typical deaths of people from illnesses such as cancer, heart disease or liver disease – begins with patients noticing a loss of strength and energy.

Activities which they used to take in their stride now require physical and mental effort. They may need afternoon naps to get through the day, with the naps becoming steadily longer or more frequent. At some stage, they find they are asleep more than they are awake. As their organs and body systems continue inexorably to fail, other changes emerge. Their weight often drops as their appetite fades. Their sense of thirst wanes too, so that although the patient eats and drinks very little, they are not, usually, particularly hungry or thirsty.

At some stage, their brain becomes so much less responsive that sleep slides into unconsciousness. As the breathing centres in the brain stem begin to shut down, a pattern may emerge of long pauses between breaths which alternate with deep, sighing breaths. This pattern, known as Cheyne-Stokes breathing, can greatly alarm those gathered at the bedside, who are convinced with each pause that their loved one has died. But when we explain what is happening, that anguish is at least in part assuaged.

Sometimes, very close to the end of life, a small amount of saliva collects at the top of a patient’s throat because they are too weak or deeply unconscious to swallow it. As breaths pass backwards and forwards through the saliva, a harsh “death rattle” is heard, causing considerable anxiety in families. Again, our explanations can be enormously helpful at alleviating some of the distress of family members.

Finally, as the patient’s heart muscle begins to fail, their blood pressure falls and the hands and feet may become cold, blue or pale. Usually there aren’t any dramas or surprises. There is a final exhalation, breath turns into air, and only gradually does it dawn on those present that the moment of death has passed.

«

unique link to this extract


Meta says it has taken down about 20 covert influence operations in 2024 • The Guardian

Robert Booth:

»

Meta has intervened to take down about 20 covert influence operations around the world this year, it has emerged – though the tech firm said fears of AI-fuelled fakery warping elections had not materialised in 2024.

Nick Clegg, the president of global affairs at the company that runs Facebook, Instagram and WhatsApp, said Russia was still the No. 1 source of the adversarial online activity but said in a briefing it was “striking” how little AI was used to try to trick voters in the busiest ever year for elections around the world.

The former British deputy prime minister revealed that Meta, which has more than 3 billion users, had to take down just over 500,000 requests to generate images on its own AI tools of Donald Trump and Kamala Harris, JD Vance and Joe Biden in the month leading up to US election day.

But the firm’s security experts had to tackle a new operation using fake accounts to manipulate public debate for a strategic goal at the rate of more than one every three weeks. The “coordinated inauthentic behaviour” incidents included a Russian network using dozens of Facebook accounts and fictitious news websites to target people in Georgia, Armenia and Azerbaijan.

Another was a Russia-based operation that employed AI to create fake news websites using brands such as Fox News and the Telegraph to try to weaken western support for Ukraine, and used Francophone fake news sites to promote Russia’s role in Africa and to criticise that of France.

“Russia remains the No 1 source of the covert influence operations we’ve disrupted to date – with 39 networks disrupted in total since 2017,” he said. The next most frequent sources of foreign interference detected by Meta are Iran and China.

Giving an evaluation of the effect of AI fakery after a wave of polls in 50 countries including the US, India, Taiwan, France, Germany and the UK, he said: “There were all sorts of warnings about the potential risks of things like widespread deepfakes and AI enabled disinformation campaigns. That’s not what we’ve seen from what we’ve monitored across our services. It seems these risks did not materialise in a significant way, and that any such impact was modest and limited in scope.”

«

The absence of AI feels like a dog that didn’t bark: what happened? Is it too early? Or not useful?
unique link to this extract


FTC bans location data company that powers the surveillance ecosystem • 404 Media

Joseph Cox:

»

The Federal Trade Commission (FTC) announced sweeping action against some of the most important companies in the location data industry on Tuesday, including those that power surveillance tools used by a wide spread of US law enforcement agencies and demanding they delete data related to certain sensitive areas like health clinics and places of worship. 

Venntel, through its parent company Gravy Analytics, takes location data from smartphones, either through ordinary apps installed on them or through the advertising ecosystem, and then provides that data feed to other companies who sell location tracking technology to the government or sells the data directly itself. Venntel is the company that provides the underlying data for a variety of other government contractors and surveillance tools, including Locate X. 404 Media and a group of other journalists recently revealed Locate X could be used to pinpoint phones that visited abortion clinics. 

The FTC says in a proposed order that Gravy and Venntel will be banned from selling, disclosing, or using sensitive location data, except in “limited circumstances” involving national security or law enforcement.

«

This is going to be interesting: how far is this cutting off going to go?
unique link to this extract


The UX of LEGO Interface Panels • Interaction Magic

George Cave:

»

Piloting an ocean exploration ship or Martian research shuttle is serious business. Let’s hope the control panel is up to scratch. Two studs wide and angled at 45°, the ubiquitous “2×2 decorated slope” is a LEGO minifigure’s interface to the world.

These iconic, low-resolution designs are the perfect tool to learn the basics of physical interface design. Armed with 52 different bricks, let’s see what they can teach us about the design, layout and organisation of complex interfaces.

Welcome to the world of LEGO UX design.

At a glance, the variety of these designs can be overwhelming, but it’s clear that some of these interfaces look far more chaotic than others. Most interfaces in our world contain a blend of digital screens and analog inputs like switches and dials. These LEGO panels are no different.

Plotting the panels across these two axes reveals a few different clusters. Screens with an accompanying row of buttons sit in the top left. A small cluster of very organised switch panels lies to the far right. The centre bottom is occupied by some wild concepts that are hard to understand, even after several glances.

Designing a complex machine interface is a juggling act of many different factors from ergonomics to engineering. But we can break down the problem into two key questions:

• How can we differentiate between the function of different inputs?
• How can we organise the many inputs and outputs so that we understand how they relate to each other?
• Let’s take a deeper look at tackling these two challenges in LEGO.

«

This is from August 2020 (hey, people had to do something during lockdown, don’t be judgey) But it makes excellent points about interface design: don’t make buttons that do different things look the same.
unique link to this extract


OpenAI explores advertising as it steps up revenue drive • Financial Times

Madhumita Murgia, Cristina Criddle and George Hammond:

»

OpenAI is discussing plans to introduce advertising to its artificial intelligence products, as the ChatGPT maker seeks new revenue sources as it restructures as a for-profit company.

Sarah Friar, chief financial officer at OpenAI, told the Financial Times in an interview that the $150bn AI start-up was weighing up an ads model, adding that it planned to be “thoughtful about when and where we implement them [ads]”.

The San Francisco-based group, which in October secured $6.6bn in new funding, has been hiring advertising talent from big tech rivals such as Meta and Google, according to multiple people familiar with the matter and an FT analysis of LinkedIn accounts.

In a statement following the interview, Friar added: “Our current business is experiencing rapid growth and we see significant opportunities within our existing business model. While we’re open to exploring other revenue streams in the future, we have no active plans to pursue advertising.”

OpenAI is stepping up efforts to generate revenue from its products, such as its AI-powered search engine, as it seeks to capitalise on its early lead in the booming AI sector. Its smaller rival Perplexity is already piloting advertising in its AI-powered search engine.

Friar, who previously held leadership roles at companies such as Nextdoor, Square and Salesforce, pointed to the wealth of advertising experience between herself and Kevin Weil, the company’s chief product officer.

«

Just bringing up another squadron of tanks to park on Google’s lawn to go alongside the search engine.
unique link to this extract


Ultra-processed news: why local journalism has gone weird • West Country Voices

Philippa Davies:

»

Something weird has happened to local news – and it led to me resigning from my job as a journalist here in Devon. So here’s the inside story.

Until recently I was employed by one of the UK’s biggest news publishers, which has hundreds of titles across the country. I was working across three local weekly print titles and their websites. We were very short-staffed, but I was keen to do my best to report on the events and issues I thought were important to our readers – both in print and online.

Although local papers have been in decline for many years – in inverse proportion to the growth of digital media – the news agenda was still more or less unchanged when I started this job six years ago. The aim was to provide accurate, well-written stories about local issues, reporting on breaking news (fires, accidents, crime etc), major planning applications, council services, politics, developments in health and education, business and employment, and so on. Alongside that were the traditional ‘local paper’ stories, of a more human-interest kind.

There was pressure to get big stories published on our websites as quickly as possible, but the news content of the print editions and the websites was pretty much the same.

Then, in the last couple of years, a dramatic change emerged, which escalated rapidly. Old-style news values were replaced by a single criterion for assessing the importance of a story: will it get a lot of online page views? Will it pull in high audience numbers to boost advertising revenue?

And this is how local journalism turned into ultra-processed news.

«

Pretty much what you’d expect – and completely depressing. Ironically, this piece by Davies has probably had more views than anything she wrote for the news publisher (unnamed) which gave her dire targets for page views per month.

The problem is, if you’re not making the money from print (though that can be done! My local market town’s local paper manages it) you’re going to struggle with local news.
unique link to this extract


Apple’s foldable iPhone is expected to save a surprisingly declining market • 9to5Mac

Ryan Christoffel:

»

A new report from DSCC highlights the struggles that foldable smartphones are facing in the market. They just endured their first year-over-year decline, and things are expected to get worse. The report notes one reason for optimism though: Apple’s upcoming foldable iPhone.

Foldables are still in their early days, but they’ve just hit an unfortunate peak. For the first time ever, Q3 2024 saw a year-over-year decline in panel procurements.

This apparently isn’t just a blip, either, as the display analysts at DSCC expect an entire year of decline coming up in 2025.

…The foldable iPhone is currently expected to launch in late 2026, and as a result DSCC forecasts that to be a record year for foldables overall as Apple reinvigorates the category:

»

Although the market has stalled and will decline for the first time on a panel procurement basis in 2025, there is reason for optimism. Apple is expected to enter the foldable market in 2H’26 and given their dominant position in flagship smartphones could generate significant growth for the foldable smartphone market. Any improvement in form factor, functionality, use cases, durability, etc. could drive new demand for this market. As a result, 2026 is expected to be a record year for foldables with over 30% growth and with over 20% growth projected for 2027 and 2028 as well.

«

«

I’ll believe it when I see it. Apple is struggling enough on the LLM front, so why it would give itself the headache of trying to launch into a stagnant niche beats me.
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

1 thought on “Start Up No.2347: OpenAI blames ‘glitch’ for Unsayable names, FTC blocks location data seller, a Verge subscription?, and more

  1. Apparently foldables still fail a lot compared to monoblock devices.

    I doubt Apple wants a new butterfly keyboard scandal. Plus it seems that there isn’t much of a market for them. I’ve seen a foldable device maybe half a dozen times in the wild.

    And I always gasp at the fold part in the display :/

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.