How do you distinguish between 'complex' and 'complicated'? How do you manage complexity? I find these to be questions that apply just as much to technical matters as to everyday life, and this article offers an excellent starting point for an answer.

Hugging Face’s post-mortem on the hack of their platform by OpenAI is fascinating—particularly the section detailing how they used LLMs themselves (which ones, why, how, etc.) and noting that the initial models employed were useless: analysis showed that requests 'were blocked by the providers' safety guardrails, which cannot distinguish an incident responder from an attacker.'

An attack that I found pretty wild, involving malware hidden in SVG files and targeting developers through fake recruitment campaigns.

This article lays the groundwork for understanding this concept, which refers to “the expansion of a system or technology beyond its original purposes”: how does it differ from innovation? Why is it particularly applicable to the realm of data sovereignty?

This article examines US labor market data for the IT sector and reached three conclusions: 1. There is indeed a significant and persistent rise in layoffs within the sector. 2. However, those laid off quickly find new jobs. 3. So far, no analysis has managed to establish a concrete causal link between the widespread adoption of AI and layoffs in the sector.

Alors, elles y sont ou pas ?

The life of a 200ms request, from the perspective of the request itself. A very well done interactive story.

Quelles sont les différentes couches dans les modèles Claude

In the 'tinker against extremism' category, discover an idea for creating your own anarchist library.

Un ensemble d'idées fortes pour repenser la responsabilité numérique : accès Internet, réduction d'impact, et critique de l’infrastructure actuelle.

Mozilla analyzes the privacy flaws of reproductive health applications: massive data collection and non-transparent sharing.

A profound critique of the open source ecosystem's dependence on GitHub and a call to regain control of the tools and infrastructure.

Rappel du rôle fondamental de l’accessibilité dans la mission originelle du web, et importance des normes WAI/WCAG pour un numérique réellement universel.

Advice for entry-level developers in a challenging job market: understanding the context, the biases of success stories, and the realities of the industry.

A defense of pen and paper for thinking about complex algorithms, a useful tool even for experienced developers.

Apprise permet d’envoyer des notifications vers la plupart des services populaires : Telegram, Slack, SNS, Gotify, etc.

Des cours reconnus sur Flexbox, Grid et Markdown, publiés chaque année avec une excellente pédagogie.

Un tutoriel clair pour créer un blog gratuit via GitHub Pages, domaine inclus — un excellent point d’entrée pour débuter.

Introduction to color fonts and their uses: emoji, multi-colored fonts, and creative implications for design.

An introduction to timing attacks: exploiting the duration of server responses to retrieve secret information.

A Pew study on the digital habits of teenagers: increased web usage, disparities across social categories.

Analyse d’une politique américaine interdisant les paywalls sur les recherches financées par les contribuables : enjeux d’accès libre et implications économiques.

All you've ever wanted to know about responsive web design: media queries, flexible grids, images, and best practices.

One day, it struck me that I haven’t had a clue why vertical rhythm was important. Two more questions quickly arose following that thought: “How does Vertical Rhythm improve the design of the site? What lessons can I draw from Vertical Rhythm so I can improve my design? "

Design is hard. But I've come to realize that laying out texts properly is 80% of what makes something look clean, and is the easiest thing you can do to make your design much nicer and more usable. "

No matter what you build—be it a next generation video editing app, an addictive word game, or a future online social networking app—you will always find yourself in need of a few basic building blocks. "

As the web has evolved, so too has the architecture for the development of these applications. There are many core architectures for building applications for the web these days. The most popular architecture employed by web developers today is the Single Page App (SPA), but we are transitioning to a new and improved architecture for building web applications.

I believe that the conventional idea of “writing a program” is headed for extinction, and indeed, for all but very specialized applications, most software, as we know it, will be replaced by AI systems that are trained rather than programmed.

Users need open source projects, but open source projects do not need users. That asymmetry is, I believe, at the crux of the open source sustainability problem.

Games industry veteran Mike Acton rattled off a sample of 50 things he expects of developers he works with.

Les logiciels de génération d’images adossés à l’intelligence artificielle (IA) sont capables d’inventer des objets en leur attribuant n’importe quel style. Nous leur avons demandé de s’inspirer de grands noms de l’architecture et du design pour imaginer ce vestiaire impossible.

We’re glad to present you our new Text-to-Music demo interface. Now as a Google Colab, and soon we’ll add this feature as a simple form on our website. This has already gone viral, so the community has questions about how everything works.

In this article, Yosra Emad explains how to create a rollercoaster path that a ball follows using cubic beziers and CSS transitions. You’ll also learn how the cubic-bezier function in CSS works in detail and how to stack multiple simple animations to create one complex one.

Fluid typography: the idea that font sizes should change according to screen size.

Fluid typography: the idea that font sizes should change according to screen size.

A collection of 70 hand-picked, web-based tools which are actually useful. Each will generate pure CSS without the need for JS or any external libraries.

Depuis au moins une vingtaine d'années, les études critiques sur les effets sociaux des méthodes informatiques de calcul - souvent appelées «algorithmes» - se sont multipliées. Les monographies de Bucher (2018), Noble (2016), O'Neil (2016) ou Steiner (2012) ont, parmi d'autres, mis en exergue les processus de discrimination et de mise en invisibilité subrepticement induits par le recours généralisé à des dispositifs algorithmiques.

What if someone on your team suddenly won the lottery? What if they signed out of the team chat, and you never heard from them again? Well, first of all, good for them, but would your team be in trouble? Would all of their knowledge go out the door as they sailed off into the sunset?

Writing notes by hand would have given me several different tangible resources that could help me find the critical missing information: a stronger memory of the meeting I was in, the gaps in the details of the discussion that occurred, and the notes themselves that would help me trigger a stronger recall of the events just by reviewing them on paper.

How has the process of technological change in the Internet era influenced the way we organize economic activities? In this chapter we discuss how information technologies foster the emergent design and user-driven design of websites and other online media...

Imagine for a moment the perfect organizational system. One that supported and enhanced the work you do, telling you exactly where to put a piece of information, and exactly where to find it when you needed it.

Les menaces évoluant quotidiennement, les “défenseurs” (mais également les auditeurs techniques / pentesters pour d’autres objectifs) doivent donc disposer des informations les plus « fraîches » possibles pour adapter leurs dispositifs de protection, de détection et de réponse, informer, convaincre, sensibiliser leurs clients ou leur direction.

D'aucuns diront que Crowdsec c'est le Waze de la cybersécurité. L'image me parait plutôt bonne, car effectivement on va s'adosser à un service d'alertes centralisées pour obtenir non pas l'emplacement de la maréchaussée mais bien une base de réputation d'ips.

What if social media wasn't engineered to serve capitalism's need for growth? How might online collective communication be different if our time and attention were treated as the limited and precious resources that they are? Minus is an experiment to ask these questions, a finite social network where users get only 100 posts—for life.

Calafou aims to be a place for inventing practices to escape capitalism, patriarchy, and forms of social oppression in general, and produces a great deal of writing to document the international meetings that take place there and the ideas that develop there. It is there that Anarchaserver, a self-managed feminist server, was born.

Pure is ridiculously tiny. The entire set of modules clocks in at 3.5KB* minified and gzipped. Crafted with mobile devices in mind, it was important to us to keep our file sizes small, and every line of CSS was carefully considered.

Dans l'idée, je m'inspire évidemment du serment d'Hippocrate. On pourra m'objecter qu'un médecin et un développeur n'ont pas vraiment les mêmes prérogatives. Pourtant, notre société ne pourrait pas fonctionner sans informatique, et sans des gens qui l'entretiennent, et développent les outils dont elle a besoin.

Almost all of the fears around AI that are circulating right now — fears of cheating, of “disinformation,” of scamming and spamming, etc. — all actually boil down to one thing: fear about what happens when a whole bunch of people, some of whom are stupid and/or irresponsible and/or malicious, instantly level up and get really good at making cultural objects.

AI art generators can then be used to create artwork that replicates the style of specific artists.Whether or not these systems infringe on copyright law is a complicated question which experts say will need to be settled in the courts.

Dans cet article nous allons vous présenter quelques outils d'IA gratuits permettant de détecter si les contenus ont été générés automatiquement via un outil d'IA tels que ChatGPT, GPT3 et GPT2.

In behavioral psychology, the mirror test is designed to discover animals' capacity for self-awareness. There are a few variations of the test, but the essence is always the same: do animals recognize themselves in the mirror or think it's another being altogether?

Want to make sure your API keys are safe and sound when working with React ?

Or said differently, exactly when and how does React update the view ? Many blog posts, conference talks, and tweet threads have been dedicated to this seemingly simple topic. And yet, for some reason, it's still a topic that even experienced React developers have some (often unkhown) misconception about.

React is powerful and we love to use it, but sometimes we struggle to deliver an optimized product.

Dans ce papier, on découvre l'algorithme d'attention, dont le potentiel a largement dépassé le cadre de la traduction. En effet, c'est l'un des principaux composants de GPT-3/4/5. Il joue un rôle essentiel dans la précision et la qualité des résultats de l'algorithme.

Entre deux aventures spatiales, l'auteur reprend son costume de médiateur numérique et nous conseille sur les bons réflexes à adopter.

page d'information écologique du site suisse Infomaniak.

Nous aborderons les trois pilliers de l'IA responsable : le monitoring, l'optimisation et le recyclage.

If there's one thing that defines the Marvel movie era, it's the post-credit scene. From the first moment Samuel L. Jackson appeared onscreen to announce the 'Avengers Initiative', our brains were rewired to wait until the screen goes black.

Who can create the worst UI/UX ? A thread with hilarious examples of bad user interfaces.

To regain the ability to server render, these frameworks also run on the server to generate HTML. We get to author and maintain a single application in a single language. When our app starts in the browser, these frameworks re-run the same code adding event handlers and ensuring the app is in the correct state. And that "re-hydration" (later shortened to hydration) is what enables the application to be interactive. Sound good so far? Well, there is a problem.

In March of 2022 Node.js got a new built-in test runner via node:test module. I have evaluated the test runner and made several presentations showing its features and comparing the new built-in test runner with the other test runners like Ava, Jest, and Mocha.

The Vercel AI SDK is a library for building edge-ready AI-powered streaming text and chat UIs

No matter how much you try to write code that's readable, and how hard everyone else around you tries to do the same, the reality of life is you will have to read code that you find difficult to understand at some point in your career.

Good questions save time. Bad questions waste time. Bad questions create unnecessary back-and-forth conversations, which create frustration and conflict. People that ask bad questions get frustrated because they can’t get help, and people that are trying to help get frustrated because answering bad questions is so damn frustrating.

L’entropie a été initialement définie pour mesurer la quantité d’information émise par un émetteur et déterminer le débit d’information de cette source. En particulier, l’entropie fourni un minimum pour la taille moyenne des informations transmise (avec moins de caractères on perdrait quelque chose).

A l’image de ce qui se produisit dans la sphère politique depuis le tout début des années 2010 – Barack Obama est élu pour la 1ère fois en 2008 -, avec le passage d’une ère du storytelling (basé entre autres sur de l’analyse de données) à une ère du clash (reposant sur une maîtrise rhétorique des discours médiatiques), c’est désormais l’ensemble de l’écosystème des discours médiatiques mais aussi d’une partie de plus en plus significative de nos interactions sociales avec l’information qui nous mène d’une société de la “data” à une société du “prompt” et du “script.

I don't think Google gives a shit about the $10m it gets from predatory fake abortion clinics. But I think the company believes that the PR trouble it would get into for blocking them – and the expense it would incur in trying to catch and block fake abortion clinic ads – are real liabilities. In other words, it's not about the $10m it would lose by blocking the ads – Google wants to avoid the political heat it would take from forced birth fanatics and cost of the human reviewers who would have to double-check rejected ads.

As Meta struggles with layoffs and its so far unrealized dreams of the metaverse, “I think Mark is also getting a feeling that he is not respected,” said Bhaskar Chakravorti, dean of global business at Tufts University’s Fletcher School. The cage match was “a way for him to show that, 'Look, he is edgy. He can be a tech bro just like the next guy.

On n'a jamais trop d'algorithmes (enfin quoique, là, on frôle l'overdose).

`<Suspense />` allows you to display a fallback while waiting for the child component to finish loading. 'Just a fancy way to render a loading state' until now, according to Elan Medoff, but, with React 18, new features increase its appeal. Or could have increased its appeal, according to the author of this article, because it's apparently a bit of a mess and depends a lot on the architecture of your React app (client-side rendering, server-side rendering, or server components).

In this article—which has generated a lot of reactions and with which I don't completely agree—Josh Collinworth attacks React itself. He lists the things that are missed when doing React and emphasizes how outdated React has become: 'React was designed seven Taylor Swift albums ago, for a world where John Mayer and Jennifer Aniston were still dating.'

The article explains well why and when this error occurs: because the pre-rendered file on the server side is not exactly found by React on the client side (for various reasons, from tag errors to capricious browsers). The solution: reconsider the logic.

Thanks to data from Datos, which analyzes web behavior on over 20 million devices worldwide, the author of the article provides a brief analysis of behaviors related to ChatGPT. While traffic increased by 900% at its launch, since May, traffic has declined by 30%. The data also allows us to know what ChatGPT is used for: primarily programming, with 29% of prompts! The charts are telling and help put ChatGPT into perspective.

The article reviews the new environmental APIs developed by Google Cloud, around pollen, air quality, etc. The idea is to provide developers with more and more tools for creating eco-friendly apps.

Aryan analyzes the SQL meme by Jordan Lewis which presents an iceberg with 7 zones, 5 of which are submerged. The author says he became aware of his SQL shortcomings upon seeing this meme, and I had exactly the same impression: fortunately, he reviews the different zones in this article which allows everyone to get up to speed on SQL and learn a lot of relatively useful things (there are 87 datatypes in PostgreSQL 14.1 for example).

Que l'on écrive du code ou une thèse, il est toujours important de se rappeler pour qui et pourquoi on écrit. Quand j'ai commencé à écrire ma thèse, cette question a été centrale : qui allait me lire, qui était susceptible de me lire, qui je voulais voir me lire... les réponses à ces questions m'ont aidée à savoir comment écrire. Je pense que ces questions méritent aussi d'étre posées quand on écrit du code et Mathieu de Gracia le rappelle en allant plus loin et en nous donnant ses critères d'un code lisible selon lui. J'en retiens un maître mot : il faut être adéquat

4 minutes to finally know when to rebase or merge, clear and concise, as always by ByteByteGo whose resources I warmly recommend (newsletter, YouTube channel, etc.)

Overview of the changes between PHP 2023 and PHP 2012, trying to convince us that PHP is ultimately the future (but can we really trust someone who quotes the Lumineers?).

A nice review of Stefano Mancuso's book, *The Nation of Plants*, which reminds us of the balance in which all species live, to the point that, as the anecdote tells us, the number of cats in a country can reverse its economic situation. This was the case with communist China, which shot itself in the foot with its campaigns to eradicate animals that carried serious diseases like plague or cholera: 'The estimates provided by the Chinese government, totally unreliable for their enormity, indicated a billion and a half rats and a billion sparrows killed. Even though they are enormously exaggerated, these figures nevertheless tell us of a massacre whose dramatic consequences would soon be evident. Sparrows, in fact, do not feed exclusively on hulled grains. On the contrary, their main food supply is insects. [...] The number of locusts began to increase exponentially, and immense swarms of insects making their way through the fields of China destroyed most of the crops.'

ChatGPT also has its little slumps, especially during winter. Soon an AI blue Monday?

To gently get started with using AI in apps.

Documents and tracks vulnerabilities of LLMs.

Researcher Gareth Heyes explains the possible vulnerabilities of CSS. Yes, CSS. You thought "they allow styles because they're just styles right? What possible damage can you do with just CSS?". He reviews known techniques as well as new ones related to the :has selector attribute.

The article reviews a known vulnerability of Google OAuth which is quite significant and that Google has not yet resolved.

A random dude who trolled early 2024 by launching his package "everything". The package depends on all other existing npm packages. Aboukhadijeh reviews this little troll, reminding that he is not the first clever person to have this idea.

'Every complex system today stands on the shoulders of lessons from earlier, formative times'. This reminds me of Raymonde Moulin who says that knowing the past 'allows us to find the roots of the present'.

Why use react query: an instructive reminder of the problems with fetch in useEffect, and ways to make states "consistent, predictable", to manage loading and errors, etc.

A nice thread to discover interesting talks.

Un projet passé dans pas mal de médias et que je suis allée voir plus en détail : Synthetic Memories, du collectif espagnol Domestic Data Streamers fondé en 2013 à Barcelone et présent dans plus de 45 pays. Le projet est assez simple : proposer de compléter - plutôt que de reconstruire - les souvenirs perdus grâce à l’IA.

A recent article that takes the question of CSS performance quite seriously - it speaks to me. Daniel Nagy decided to take matters into his own hands through an experiment: migrating his website to CSS to ultimately have 3 different versions: one version with CSS in a separate file (2000 lines!); one version with inline CSS (e.g., `<div style="color: blue;" />`); a final version writing CSS in a `<style>` tag in each component.

Le procédé le plus classique et le plus utilisé actuellement serait celui du code dans des images. Dans cet article, Claudio Cortin revient sur les techniques de steganographie et notamment celle du LSB (less significant bit) : les images sont des pixels, et chaque pixel est composée de R, G et B et chaque R-G-B est un octet (8 bit). Mais si on change le dernier bit de chaque R-G-B, la couleur est quasiment identique… On peut donc utiliser ce bit pour y stocker une autre info.

I am following with interest the work of the media 'Rest of the World' which has as its editorial angle tech monitoring everywhere except in the West. The newsletter is full of links and references to articles that give an overview of how tech works elsewhere. The job market for developers, for example, is still and regularly - even if this view has become more nuanced recently - presented as a rather flourishing market and not (too) restrictive in the sense that the shortage of developers forces their employers to be a little more flexible than in more saturated markets... This article shows us that no: the developer market is radically different in other parts of the world.

It's just funny (and comforting if you feel like your work doesn't make sense).

Lots of links to 404 pages, not necessarily fun, but always interesting. It reminded me of the project of my former colleague Omer Pesquer who lists the best 404 pages of museums.

For example, if you have trouble really understanding what a “react server component” is, Paul Scanlon gives a good explanation based on his small experiments done on Waku, a minimalist React framework.

Thank you Andrew Israel, because I was really wondering if I was getting dumber. The article reviews the Nextjs app router and its drawbacks, which sometimes lead to quite convoluted setups to manage the aforementioned server components.

Sam Selikoff encourages us to think of React as a programming language for UI, with its own rules and syntax, thanks to a few additional examples of syntax choices specific to React and the advantages/disadvantages they can bring.

An article that is both fascinating and distressing about how Sam Curry realized that his modem was hacked. It took him several years, and the story is still not very clear, but it is interesting from a technical point of view: Sam Curry explains well the technical investigations - or less technical ones, like going to his operator to ask for permission to dismantle the modem he was renting - to understand what is happening to his modem.

Personally not a Windows user, I was not aware of one of their latest features called "Recall” which involves taking snapshots of your screen to create a large database in which you can perform intelligent searches. However, as Kevin points out, a large majority of Windows users use their PC to “play games, watch porn, and live their lives as human beings who make mistakes..

Une vidéo franchement angoissante sur les failles de notre système téléphonique mais extrêmement instructive sur comment ça fonctionne en fait, aujourd'hui, un appel téléphonique. Vous savez, vous ? La vidéo remonte dans l'histoire en montrant comment notre système téléphonique peut être hacké - ce qui avait déjà été fait par Steve Jobs et Steve Wozniak - et comment il est possible de le faire aujourd'hui : recevoir vos appels téléphoniques ? Possible, et même pas besoin d'être un mastermind du hacking.

Apart from being an English word with a properly indecent number of vowels, “queueing” can be done in different ways and with different objectives, as different types of queues do not have the same performance. Sam Rose, in addition to explaining it incredibly well, has provided clear and impactful animations in his article.

Promises in JS are as fundamental as they are misunderstood. Indeed, they are a fundamental part of JS, and to understand how a promise works and master its usage, you need to REALLY understand how JavaScript works: what does it mean to be single-threaded? What is a callback and how does it work? Josh Comeau revisits these concepts with well-thought-out interactive examples.

Another article by Sam Rose, this time based on the work of Colin Scott, who compiled some figures that are always good to know for developers: the latency of the most important operations. How long does it take to read 1MB sequentially? The big plus of this visualization: we can see the evolution of these operations over the years. In your opinion, what will take less time in 2024 than in 2020: reading 1MB sequentially or 4K randomly?

Because sometimes, we spell it wrong.

For a dive into the history of web design.

ou le capitalisme en action.

Doug Turnbull revisits the fact that juniors will disappear from the developer job market, replaced by AI. For him, this completely misses why we have juniors in companies: they are not “code monkeys” who are there to churn out code.

A book I discovered while revisiting the work Allan Fisher did at Carnegie University within the computer science department. I didn't know he had written a book, full of testimonials.

A small case study on the use of the CSS content-visibility property, which allows ignoring the rendering of an element. If the content is off-screen, it improves the page loading time.

A really comprehensive article to wrap up all the things to consider when talking about web performance. After recalling why web performance is important (the numbers are still a bit scary) and how to measure it (spoiler: with tools made for it), the article identifies 7 action points to improve performance: html, css, javascript, images, fonts, videos, and hosting.

A 178GB repo, 2500 packages, and 20 million lines of code: that was Jonathan's problem at Microsoft. He draws 3 lessons from it. Spoiler, in the end, the repo is 5GB.

Un épisode de If this then dev qui plonge dans les algorithmes de recommandations avec un interview d’Arielle Marouani, data scientist à Deezer. Il creuse sur comment nos gouts musicaux sont “prédits” par les algorithmes de recommandations. Hyper intéressant pour comprendre ce que c’est une matrice, comment on décompose et pour se rendre compte de tout ce qui est aggrégé en matière de data pour pouvoir calculer tout ca (en gros: tout.

On Spotify's tech blog, there's also a lot of information about their recommendation and personalization algorithms, and I came across an article I found interesting because it addresses a problem with recommendations: how do you know if data is 'relevant' or not? In Spotify's case, how do you know if having 200 hours of Frozen listening is really relevant for building your weekly playlist (or if, by any chance, you just spent the week with your nephew and were a bit short on ideas)? 'For a long time, that personalization has struggled to accurately reflect certain users' music preferences because of one main factor: our systems have treated everything a listener streams as 'representative' of their musical taste.' Spoiler alert: they haven't solved it by processing their data differently, but by giving you the responsibility to do so with their new 'Exclude from your taste profile' feature.

I loved this article from the Meilisearch blog because it really helped me understand the differences between personalization and content recommendation, with concrete examples from major platforms (Reddit ‘Because you’ve visited this’ vs Netflix “you may like this”), a typology of different ways to recommend, and some non-jargon technical aspects. At a time when personalized content or AI recommendations are talked about everywhere, I found this article very helpful to navigate this jungle.

I find that Eduardo Boucas addresses a crucial topic when choosing a technology stack: the governance questions of that stack. The example he takes of Nextjs is really interesting, as it is like many tools, an open-source framework, but developed, maintained, etc. by Vercel, a completely for-profit company. A bit like React with Meta. He shows through concrete examples what problems this can pose or at least the questions it raises.

An article by Jason Chan, former security lead at Netflix, who goes somewhat against the mantra that all developers should care about security when they code and be "security first." For Jason Chan, an engineer should focus on what they know how to do and what they were hired for: if it's data, it's not security.

Dmytro explains simply and quickly the limits of encryption and how it can give clues about its content. It's security but not semantic security, which does not allow guessing the content of a message from its form.

Access control is a fundamental concept in security. With this article from the Zero Trust blog, the concepts of whitelisting (no one enters except…) or blacklisting (everyone enters except…) become clear and, above all, are illustrated with examples and concrete cases of their application.

Aurora Scharff, a great React specialist, gives us a significant hand in understanding how to properly articulate server components and client components with concrete examples: how to do a “show more” for example?

One of the best descriptions of React, written by an AI for real developers.

A comprehensive guide to getting started with prompt engineering on AI as a developer. With categories of prompt types and techniques depending on the desired results (constraint anchoring, output/input, etc.), concrete examples, and ready-to-use prompts to test. A must-have.

It all starts with a simple calculation asked to several LLMs in spring 2025 and a unanimous answer: 8.8 - 8.11 = ?. So why do most language models fail to do the calculations correctly? Damon Segal explores this mystery and teaches us a lot about AI...

Why relying on the number of stars of a GitHub repo is not so smart? Eugen Rojavski explains how rating systems can be manipulated.

The co-founder of Firefox tells how he realized that some users just want to see the world burn and takes the opportunity to review how we design the security of our applications.

In November 2025, a second Shai-Hulud attack compromised more than 300 NPM packages. A look back at this security flaw.

Do you know what OSINT is? Open Source Intelligence is the way of gathering information using only open source data. Nathaniel Fried takes the example of Luigi Mangione to show us how easy it is to collect an impressive amount of information about a person using only public data.

AIs still fail at many things and we discover why reading the time on a clock and creating an image of a clock is so difficult for AIs.

In the first article on the Mozilla Foundation's new blog, the authors explain the concept of the post-naive internet: where the utopia of the internet is no longer to be for everyone, but specifically for certain communities.

On the Internet, you have to consent all the time: ticking hundreds of 'I agree' boxes. And yet, do we really consent? Do we have a choice?

The number of mobile applications has exploded: today, there is an app for everything. The article reminds us why this is not a good idea and the major difference between a mobile application and a website.

A good infographic overview of the race for AI supremacy between countries and major players in the field.

Among the limitations of AI, there is apparently the inability to know which emojis really exist. Does the seahorse exist or not? Why do AIs hallucinate?