Start Up No.1964: TikTok’s widespread data sources, Twitter devs face subscription crunch, Snapchat adds AI bot, and more

In Germany, Wirecard seemed to be a huge success story – but there was a huge hole behind the facade. CC-licensed photo by Web Summit 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.

The Overspill is going on a break for two weeks. See you next time on Monday 20 March.

It’s Friday, so there’s another post due at the Social Warming Substack at about 0845 UK time. It’s about Scott Adams.

A selection of 10 links for you. Use them wisely. I’m @charlesarthur on Twitter. On Mastodon: Observations and links welcome.

We found 28,000 apps sending TikTok data. Banning the app won’t help • Gizmodo

Thomas Germain:


Joe Biden gave federal agencies 30 days to remove TikTok from government devices earlier this week. Until now, most politicians intent on punishing TikTok have focused solely on banning the app itself, but, according to a memo reviewed by Reuters, federal agencies must also “prohibit internet traffic from reaching the company.” That’s a lot more complicated than it sounds. Gizmodo has learned that tens of thousands of apps—many which may already be installed on federal employees’ work phones—use code that sends data to TikTok.

Some 28,251 apps use TikTok’s software development kits, (SDKs), tools which integrates apps with TikTok’s systems—and send TikTok user data—for functions like ads within TikTok, logging in, and sharing videos from the app. That’s according to a search conducted by Gizmodo and corroborated by AppFigures, an analytics company. But apps aren’t TikTok’s only source of data. There are TikTok trackers spread across even more websites. The type of data sharing TikTok is doing is just as common on other parts of the internet.

The apps using the TikTok SDK include popular games like Mobile Legends: Bang Bang, Trivia Crack, and Fruit Ninja, photo editors like VSCO and Canva, lesser-known dating apps, weather apps, WiFi utilities, and a wide variety of other apps in nearly every category. The developers for the apps listed above did not immediately respond to a request for comment.

“A simple ban on the TikTok app itself is not going to stop data flowing to TikTok,” said Daniel Kahn Gillmor, a senior staff technologist at the American Civil Liberties Union. “TikTok has software in other places, not to mention TikTok trackers spread across other parts of the web. I don’t have a TikTok account, but there are still plenty of ways the company can get data about me.”


unique link to this extract

Tweetbot and Twitterrific users can support the developers by declining subscription refunds • MacRumors

Juli Clover:


Tweetbot and Twitterrific, two of the most used Twitter clients, had subscription offerings and thousands of customers that paid for subscriptions on a yearly basis. With the apps unable to function, pro-rated refunds are set to be automatically issued to subscribers next month, which will heavily impact businesses that had no warning their income stream would be cut off.

Those refunds are going to be paid largely by Tweetbot and Twitterific rather than Apple. As John Gruber points out on Daring Fireball, this is akin to a person getting fired and then having to pay back their last six months of salary. It is a significant financial blow to app developers put out of business by Twitter’s snap decision.

Tweetbot and Twitterrific have teamed up to offer multiple options to customers who are due refunds, and customers who want to help need to do the following:

• Open Tweetbot or Twitterrific (or redownload the apps if they’ve been deleted and open them).
• Choose the “I don’t need a refund button.” Alternatively, for Tweetbot, choose to transfer the subscription over to the new Ivory app for Mastodon.

Because refunds are being issued automatically, Tweetbot and Twitterrific customers who have been happy with their service and want to help the developers out will have to manually opt out using this method.


Please do this if you subscribed to either app.
unique link to this extract

Honestly, it’s probably the phones • Noahpinion

Noah Smith on the argument about teen unhappiness in the US:


The first reason smartphones should be our prior is that the timing just lines up really well. The smartphone was invented in 2007, but it didn’t really become commonplace until the 2010s, exactly when teen happiness fell off a cliff.

Younger Americans adopted the technology more quickly than older ones; 2010-11 seems to have been an especially important moment. And of course the “killer app” for smartphones was social media. When you had to go to a computer to check Facebook or Twitter, you could only experience it intermittently; now, with a smartphone in your pocket and notifications enabled, you were on every app all the time.

Why would that make us unhappy? There’s an obvious reason: social isolation.

Pretty much everyone knows that social isolation makes people less happy, and research strongly backs this up. It’s known to be a suicide risk. The worst punishment in a prison is solitary confinement, which some view as a form of torture. In case you doubt that the relationship between social isolation and unhappiness is causal, you should recall that we recently ran a gigantic natural experiment on much of society in the form of Covid, and the results were clearly negative.

But why would devices that make people more connected lead to social isolation? Isn’t that backwards? Doesn’t having access to all of their friends and acquaintances at all times via a device in their pockets mean that kids are less isolated than before?

Well, no. As the natural experiment of the pandemic demonstrated, physical interaction is important. Text is a highly attenuated medium — it’s slow and cumbersome, and an ocean of nuance and tone and emotion is lost. Even video chat is a highly incomplete substitute for physical interaction. A phone doesn’t allow you to experience the nearby physical presence of another living, breathing body — something that we spent untold eons evolving to be accustomed to. And of course that’s even before mentioning activities like sex that are far better when physical contact is involved.


More and more I think I should publish my missing chapter about exactly this chapter as an Amazon Kindle special. (Advice welcomed.)

unique link to this extract

How the biggest fraud in German history unravelled • The New Yorker

Ben Taub:


on June 18, 2020, [fintech company] Wirecard announced that nearly two billion euros was missing from the company’s accounts. The sum amounted to all the profits that Wirecard had ever reported as a public company. There were only two possibilities: the money had been stolen, or it had never existed.

The Wirecard board placed [Austrian bank executive and COO of Wirecard, Jan] Marsalek on temporary leave. The missing funds had supposedly been parked in two banks in the Philippines, and Wirecard’s Asia operations were under Marsalek’s purview. Before leaving the office that day, he told people that he was going to Manila, to track down the money.

That night, Marsalek met a friend, Martin Weiss, for pizza in Munich. Until recently, Weiss had served as the head of operations for Austria’s intelligence agency; now he trafficked in information at the intersection of politics, finance, and crime. Weiss called a far-right former Austrian parliamentarian and asked him to arrange a private jet for Marsalek, leaving from a small airfield near Vienna. The next day, another former Austrian intelligence officer allegedly drove Marsalek some two hundred and fifty miles east. Marsalek arrived at the Bad Vöslau airfield just before 8 p.m. He carried only hand luggage, paid the pilots nearly eight thousand euros in cash, and declined to take a receipt.

Philippine immigration records show that Jan Marsalek entered the country four days later, on June 23rd. But, like almost everything about Wirecard, the records had been faked. Although Austrians generally aren’t allowed dual citizenship, Marsalek held at least eight passports, including diplomatic cover from the tiny Caribbean nation of Grenada. His departure from Bad Vöslau is the last instance in which he is known to have used his real name.

The rise of Wirecard did not occur in a vacuum. Rather, it reflected a convergence of factors that made the past half decade “the golden age of fraud,” as the hedge-fund manager Jim Chanos has put it.


This is the very wildest tale, involving spying on journalists that goes miles beyond the pale.
unique link to this extract

Snapchat releases ‘My AI’ chatbot powered by ChatGPT • The Verge

Alex Heath:


Named “My AI,” Snapchat’s bot will be pinned to the app’s chat tab above conversations with friends. While initially only available for $3.99 a month Snapchat Plus subscribers, the goal is to eventually make the bot available to all of Snapchat’s 750 million monthly users, Spiegel tells The Verge.

“The big idea is that in addition to talking to our friends and family every day, we’re going to talk to AI every day,” he says. “And this is something we’re well positioned to do as a messaging service.”

At launch, My AI is essentially just a fast mobile-friendly version of ChatGPT inside Snapchat. The main difference is that Snap’s version is more restricted in what it can answer. Snap’s employees have trained it to adhere to the company’s trust and safety guidelines and not give responses that include swearing, violence, sexually explicit content, or opinions about dicey topics like politics. 

It has also been stripped of functionality that has already gotten ChatGPT banned in some schools; I tried getting it to write academic essays about various topics, for example, and it politely declined. Snap plans to keep tuning My AI as more people use it and report inappropriate answers. (I wasn’t able to conjure any in my own testing, though I’m sure others will.)

After trying My AI, it’s clear that Snap doesn’t feel the need to even explain the phenomenon that is ChatGPT, which is a testament to OpenAI building the fastest-growing consumer software product in history. Unlike OpenAI’s own ChatGPT interface, I wasn’t shown any tips or guardrails for interacting with Snap’s My AI. It opens to a blank chat page, waiting for a conversation to start.


unique link to this extract

CNET is doing big layoffs just weeks after AI-generated stories came to light • The Verge

Mia Sato:


Just weeks after news broke that tech site CNET was quietly using artificial intelligence to produce articles, the company is doing extensive layoffs that include several longtime employees, according to multiple people with knowledge of the situation. The layoffs total around a dozen people, a CNET staffer says, or about 10% of the public masthead.

The layoffs began Thursday morning and were announced internally via email by Red Ventures, the private equity-backed marketing-turned-media company that bought CNET in 2020. In the email, a Red Ventures executive suggested the cuts were made to focus CNET on areas where the site can succeed at bringing in traffic on Google search — a top priority for the company.

…Under Red Ventures, former CNET employees say the venerated publication’s focus increasingly became winning Google searches by prioritizing SEO. On these highly trafficked articles, the company crams in lucrative affiliate marketing ads for things like loans or credit cards, cashing in every time a reader signs up.

In the email, [president of financial services and the CNET group at Red Ventures, Carlos] Angrisano said CNET would focus on consumer technology, home and wellness, energy, broadband, and personal finance — the sections Red Ventures could best monetize, a current staffer says.

“But those sections are shadows of what they once were, particularly home,” the staffer says. “If you want to do that section the right way, you don’t sell off your Smart Home, get rid of its video team and cripple your editorial staff.”


Tells you everything that Angrisano is in charge of both financial services and CNET. And which comes first. Only a matter of time before CNET is shut down or sold off again; this situation won’t improve.
unique link to this extract

ChatGPT and Whisper APIs debut, allowing devs to integrate them into apps • Ars Technica

Benj Edwards:


On Wednesday, OpenAI announced the availability of developer APIs for its popular ChatGPT and Whisper AI models that will let developers integrate them into their apps. An API (application programming interface) is a set of protocols that allows different computer programs to communicate with each other. In this case, app developers can extend their apps’ abilities with OpenAI technology for an ongoing fee based on usage.

Introduced in late November, ChatGPT generates coherent text in many styles. Whisper, a speech-to-text model that launched in September, can transcribe spoken audio into text.

In particular, demand for a ChatGPT API has been huge, which led to the creation of an unauthorized API late last year that violated OpenAI’s terms of service. Now, OpenAI has introduced its own API offering to meet the demand. Compute for the APIs will happen off-device and in the cloud.

OpenAI calls its new ChatGPT API model “gpt-3.5-turbo,” which supersedes its previous “best” LLM API, “text-davinci-003.” It is priced at $0.002 per 1,000 tokens (about 750 words), which OpenAI says is about 10 times cheaper than its existing GPT-3.5 models. “Through a series of system-wide optimizations, we’ve achieved 90% cost reduction for ChatGPT since December,” writes OpenAI on its API announcement page.


That’s quite the price drop, very rapidly.
unique link to this extract

OpenAI rival Aleph Alpha is in talks with investors over major funding • Business Insider

Callum Burroughs:


Aleph Alpha, a German generative AI startup, is in talks with investors over a new funding round, Business Insider has learned.

The startup, based in Heidelberg, Germany, is talking to a number of top-tier VC investors over a round that could be as much as $100m, four sources familiar with the matter told Insider.

It comes amid a wave of investor hype in the AI market after US startup OpenAI released ChatGPT, based on its GPT-3.5 language model which was trained on Azure, to the public in November.

Industry rivals quickly saw chatbots’ potential to transform online search, with Microsoft releasing its AI-powered Bing and Google internally testing its version, Bard.

Founded in 2019 by CEO Jonas Andrulis, a former machine-learning engineer at Apple, and Samuel Weinbach, Aleph Alpha researches and develops AI systems with a focus on enterprise customers.

Talks over its raise are thought to be at an early stage with term sheets set to be submitted soon.


unique link to this extract

Can publishing survive the oncoming AI storm? • Word Count

Suw Charman-Anderson:


It should surprise nobody that there’s now a boom in LLM-created books on Amazon, although its true extent is impossible to measure as there’s no requirement to flag LLM content in book metadata or descriptions, and quite a big incentive not to. Reuters’ Greg Bensinger writes: 


Now ChatGPT appears ready to upend the staid book industry as would-be novelists and self-help gurus looking to make a quick buck are turning to the software to help create bot-made e-books and publish them through Amazon’s Kindle Direct Publishing arm. Illustrated children’s books are a favorite for such first-time authors. On YouTube, TikTok and Reddit hundreds of tutorials have spring [sic] up, demonstrating how to make a book in just a few hours. Subjects include get-rich-quick schemes, dieting advice, software coding tips and recipes.


Bensinger quotes the Authors Guild’s Mary Rasenberger, who says, “This is something we really need to be worried about, these books will flood the market and a lot of authors are going to be out of work.” Yes. Yes they will. 

Books with small amounts of text are an obvious target – they’re easy to generate on an LLM and it’s easier to keep on top of things like plot and consistency. A children’s picture book only has between 500 and 1000 words, whilst a chapter book for ages 5 to 7 will have around 5,000 to 10,000 words. With a little coaxing, an LLM is perfectly capable of producing this amount of text in a very short space of time. You can then use Dall E, MidJourney and other image creation engines to provide the images. 

These books won’t be good – this LLM-written article on how to write a book in three days using LLMs shows just how bad a whole book of this stuff can be – but that doesn’t matter, as I’ll come on to later. 

Once there’s a strategy for creating 10,000 word chapter books, it’s easy enough to extend that to 15,000 or 20,000 word novellas, at which point LLMs collide head-on with an existing trend.


unique link to this extract

Meta’s metaverse: on this evidence, the future is a bleak, cumbersome nightmare • The New European

James Ball:


Each time I took the headset off, it turned itself into sleep mode, but each time I went to the laptop for help, it complained I’d left my safe zone, so I spent merry hours reading help pages, running to my safe zone, trying to remember a code that I am then supposed to enter by firing pretend lasers on to a floating virtual keyboard. With each failed attempt, I feel another small piece of my soul slide away, never to return.

The world turns, continental plates drift, years turn into aeons, and eventually I actually enter Horizon Worlds. I even convince it that as an “experienced gamer” I can be trusted to move myself around using a joystick on the controller, rather than teleporting about (recommended to new users). A loading screen showing bright characters, unicorns and more floating atop a serene lake is almost pretty, even if the graphics look like they’re from the PS2 era.

A mini “welcome world” is almost fun. I successfully pick up and throw a paper plane after a mere nine attempts. The next activity, using the controllers to shoot rings in the air, is quite fun too – turning your head to see stuff gets quite immersive.

Of course, immersive cuts both ways. Having turned myself to see where I should head next, I forget that while things with your arms are done by moving your arms, legs are different – legs are the joystick. I reflexively step forwards before realising my error, and trip over my coffee table. Another triumph.

That one is probably my fault, and the “welcome world” was almost momentarily fun. I try to have a better attitude – and then step into the weird, empty world of Horizon Worlds. Whenever I visit, there is almost no one there. I go to a game arcade and anything I try features me, solo, playing a game that would’ve been called dated in 1993.


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

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

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