Hello again,

Hope you're having a wonderful week.

We're shaking things up this week by forgoing a top story. If you want something to get your teeth into though, the New Yorker has a tell-all on OpenAI CEO Sam Altman and how he's a big-ol' liar*, which has been taking up a considerable amount of airtime.

For this week's LT3 though, we're taking a quick break from the new format and instead we're going to quick-fire some tech news.

Let's dig in...

What is happening in the world of tech this week?

💰 OpenAI Says It’s Already Made $100 Million by Stuffing ChatGPT With Ads

It's been about 2 months since OpenAI started putting adverts into some versions of ChatGPT. In that time, it looks like they've made somewhere in the region of $100 million.

Naturally, that OpenAI puts ads in is very contentious. We seem to have grown accustomed to it when it comes from Google, and tucked over to the side. But having it embedded into the answers from our LLM is a lot.

In related news, Microsoft recently rolled out a change on its version of Copilot in GitHub, which included ads in the pull requests - system messages when users made changes to their code. The backlash was so severe, universal, and practically instant, that Microsoft backpedaled pretty quickly, claiming that it wasn't trying to advertise anything, merely including "product tips".

🔒 Brussels says EU age verification check ready amid child safety push

With fancy words like "zero-knowledge" and "cryptography" being thrown around, it's pretty clear the EU is proud of its new approach to the problem of age-verification online.

In brief, the way it works is: you download the government app, you upload your government ID, once that ID is verified a "token" is saved to your device to say "yes, this person is over 18". Then when you try to access age-restricted content from that device, rather than sharing your ID details, the token can be used as pre-approved proof that you are valid.

To be fair, on the face of it this does look a whole lot better than the UK's approach, which has been to let private companies duke it out to see who gets to hold the most amount of your private data possible. (We talked about the problems with this back in LT3 #19). The logic behind the EU's approach is that a) the only people holding your ID data is the government itself, and b) the app is open source, meaning anyone with the right technical know-how can examine the code to see exactly what it does.

But that doesn't mean this is a flawless plan. It is still causing some concern among civil liberties organisations, including the Electronic Frontier Foundation.

The app is still to be official launched, so watch this space.

🚲 Skoda develops bike bell that can bypass noise-cancelling headphones, and then gives the technology away for free

With more and more pedestrians roaming the streets with noise-cancelling headphones on, pedestrian accidents have gone up 30% in recent years. To combat this, Skoda and the University of Salford have created a new bike bell.

Noise-cancelling headphones work by basically listening to everything around you, and then playing that noise back to you, but inverted to cancel it all out as it hits your ear. In what is genuinely clever tech, this bell rings twice; once at a specific frequency to cut through the noise-cancelling, and then again in a more "traditional" bell noise while the noise-cancelling is still trying to recover.

Perhaps the best bit? The team behind the bell have opted to make the patent free, open-sourcing the technology for anyone to use and adapt. They join the ranks of fellow car maker Volvo, who in 1959 perfected the three-point safety belt, and then gave it to the world.

What's that? History and tech in one newsletter? Goodness, I spoil you.

🐟 The scientists turning recycled ocean plastic waste into roads

In a reminder that technology doesn't need to have electricity or internet running through it to be technology, scientists are trialing a new way to reuse our old plastic.

A team of marine biologists in Hawaii are exploring ways that specific types of plastic might be mixed with more traditional asphalt ingredients to make roads that - one year on - are just as durable as regular roads. And more interestingly, these plastic roads keep all their horrid little microplastics locked in.

Is this article from CBBC's Newsround? Yes. Does that make it any less important? No. Besides, look at where Lizo Mzimba's career is now. Don't knock Newsround.

💾 Website backup crippled by 1.6MB Friends GIF that was replicated 246,173 times

And to end on, no one told Discourse files were gonna save that way 👏👏👏👏 (sorry, not sorry).

Discourse is an online discussion platform, used by over 22,000 communities across the globe. This week, they ran into something of an issue when a file security policy wasn't properly thought through.

The problem came from a feature of Discourse. When a file is copied from one state to another (e.g. from a private chat to a public post), a new copy of that file is created, with its own new identifier, rather than having a single copy of the file and just sharing a link. It's a security / data privacy feature, and there are plenty of good reasons why you should do things that way.

But, if everyone's sharing the same GIF (a file that doesn't really need to be all that secure), doesn't that mean there are potentially thousands of copies of the same bit of information? Yes, dear reader, it does. And when you're looking to backup the system, you probably don't want to backup the same information 246,173 times. Luckily, there's a nerd fix for this. "Hardlinks" are a way of using one file, but linking to it multiple times within a backup. A neat, elegant solution. Could they be anymore slick?

Turns out though, there's a limit to the number of hardlinks you can assign to a file. And that limit was a little under half of how many times the GIF was shared. Happily, the team eventually solved the problem.

That's it for another week of LT3!

Next week, it's back to our regularly scheduled programming, with top stories, quick-fire news, and a blog spotlight.

If you're reading this because you've seen it on socials, or someone's shared it with you, you can get this straight into your inbox every week by subscribing.

If you don't want more emails clogging up your inbox, and are a fan of RSS (and you should be a fan of RSS), you can subscribe by pointing your reader to https://buttondown.com/willrc/rss/

I'll see you back here next week. Probably on Thursday.

Will

* allegedly, I suppose I should add. even if the evidence is pretty compelling and has been around for years. Allegedly compelling...