Let's talk tech Thursday #2
Welcome back to Let’s talk tech Thursday!
In this second edition we discuss security vs privacy, AI vs AI, and the potential comeback of an old technology.
I’m also adding a new section (yes, only on newsletter #2 and I’m already innovating!) with other links I think you might find interesting. Little in the way of commentary from me on these, other than go and look!
Let’s dig in...
Top stories
The Trump Administration Accidentally Texted Me Its War Plans
Summary
In a news story that was so ridiculous even it's author Jeffrey Goldberg didn’t believe it when it was happening, the Trump administration not only used unapproved messaging app Signal to share sensitive information about a live military operation, they also invited a journalist into the chat.
So What?
Look, I had to include this, as probably one of the most insane tech-based stories I've read in a decade. But while everyone - including me in a second - talks about the absurdity of using Signal for classified conversations, it feels appropriate to remember that there is an actual humanitarian crisis, and a decade-old civil war, that underpins all of this. Don't let the shiny-tech story distract from other stories, I guess is my point.
That said, there are some technology points I wanted to make here:
- A valid question you might have is: "Why is it a problem that Signal is being used? I heard it was one of the most secure messaging apps on the market!" While it’s true that Signal is more privacy-minded than, for example, WhatsApp, there is a very real difference between “security” and “privacy”. Signal uses similar “end-to-end encryption” technology as WhatsApp (and iMessages, for those who remember the Apple vs UK Government story from last week). But Signal is much more careful about how it handles all the stuff around your message. In the right hands this "metadata", which can include things like the device you're messing from, or your location, can be almost as revealing as the messages themselves.
- If that makes it sounds like the Department of Defense was smart to use Signal, then you're forgetting that government communications aren't supposed to be private. Or rather, they are supposed to be available when necessary for scrutiny and audit. One of the features of Signal is the ability to set messages to delete themselves after a period of time, which members of the group chat appear to have used.
- There is also a wider point here, which is that the security Signal offers only really applies to the messages. Ignoring the the US Government for a moment, I see all too often people relying on the apparent security of their messaging app, but their phone PIN is "1234". Or they put a privacy screen on their laptop, but connect to an unsecured public WiFi in their local coffee shop. There’s a lot to learn from the calamitous mistakes made, but perhaps chief among them is that neither security nor privacy are things you can completely outsource to technology. You have to play an active role.
Oh yeah, and all of that ignores the fact that they just invited a journalist to the group chat.
Related article: CNET on the differences between WhatsApp, Signal, and Telegram
Cloudflare Turns AI Against Itself With Endless Maze of Irrelevant Facts
Summary
AI is famous for stealing Intellectual Property from people, even when explicitly asked not to. To combat this, Cloudflare has launched a new feature called "AI Labyrinth", which sends AI scrapers down a maze of irrelevant (but factually accurate) AI generated content. This approach wastes the resources of crawlers while keeping the real website content protected.
So What?
It is rare to read a story where AI is the good guy when it comes to IP. This approach from Cloudflare seems at first glance quite innovative, not only stopping crawlers from stealing, but helping to swing the dial on Large Language Models back towards factual information.
If you're skeptical though, I sympathise. Ignoring the technical questions of how long before the AI crawlers catch on, there is another consideration. All of this AI needs power to generate the content, and then resources to store it. While this might not be the "AI slop" we're used to seeing, and I actually really like the idea of flooding large language my models with "good" information, it is still AI generated content that no one really asked for. The very point of this approach is to "waste the crawler's resources", presumably to cause a financial penalty on the part of the IP stealer, but at the same time almost by design causing more damage to the environment.
Also, and this seems trite to mention at this point, but I'm going to anyway, what we're talking about here is robots fighting robots. Is no one else worried about that part...?
Why we need "Shortwave 2.0"
Summary
When was the last time you thought about shortwave radio? Yeah me neither... But there might yet be life in an old technology. With the internet increasingly being censored by governments, shortwave radio could become a valuable tool for delivering information.
So what?
Originally conceived as a free and open place to share knowledge and ideas, the Internet of today has become a difficult place to navigate. More than that, it can be so easily censored. And not even censored in the traditional, authoritarian, Chinese Great Firewall kind of way. We've long known that the companies are not above manipulating what information we can and can't see online.
I wrote an article a little while ago about the resurgence of older technologies that are in many ways immune to manipulation. I was talking about RSS, but this piece looks at some of the novel ways radio can be used to deliver content in a way that can't be tampered with.
And in case you were thinking "I don't want some stuffy old people talking at me", firstly check that unconscious age-bias my friend. But secondly, you can actually use radio waves to send more than just voice. Text and even images can be sent over radio waves.
Right now, this seems to be the plaything of nerds (yes, I have been looking at how much transmitters cost). But keep an eye on this. In parts of the world where access to unfettered internet access is not a given, or in places where authoritarian technocrats cosplaying as Tony Stark keep threatening to turn off the internet, this new/old way of keeping people informed might just be vital in giving real, accurate, and up to date information to the people most in need of it.
What else is worth reading?
Don’t Use Fake Bold or Italic in Social Media
Everyone wants to stand out on the web. One of the easiest ways to do that is to jazz up your Instagram captions or your Bluesky posts with letters that mimic italic and/or bold text. You shouldn't though. As Adrian Roselli explains, complete with demos, it can have terrible consequences for those using screen readers and other accessibility tools.
Exploring AI in Refugee Support: What Does It Mean for Our Mission?
I'm a huge fan of the work CAST does. In a world where it's easy to focus on the efficiency gains technology like AI can give you, the ethics discussions sometimes get left behind. In the latest installment of a series exploring the practicalities of AI, CAST Programme Lead Ellen Smyth talks through some of the learnings from a recent workshop around the opportunities and implications of AI in refugee support.
How a battery breakthrough from China’s BYD could be about to bury Tesla
In what could be big news for the world of electric vehicles, Chinese firm BYD have created a battery and charging system that is considerably faster to charge, and give much greater range, than the current generation. BYD isn't massive in the car game outside of China (though some of London's electric busses are theirs), but with Tesla sales plummeting across Europe there could be a shake-up on the way. By some measures, it might already be here.
That's it for this edition of Let's talk tech Thursday. Thanks so much for reading.
I would love to hear from you! What do you think of the newsletter? Any thoughts on the stories above? Hit reply and let me know!
In the meantime, hope you have a great week, and see you next time,
Will