Why accepting the term AI is already a surrender. From Dartmouth 1955 to Jacques Perret's discarded synthétiseur, from the potty to Plato's cave: the name is the first correspondence, and we have been calling the machine by the wrong name for seventy years.
Why your LLM can be perfectly consistent and entirely wrong. From email interoperability to Zeno's arrow, from Socrates to Surface Stations: the battle over whether the hood opens is the oldest philosophical war in history, fought now on new ground.
From the command line to the GUI to the AI chatbot: how the opacity of our interfaces turned the tool into an invisible leash — and why transparency of purpose is the only way out.
The real danger of AI is not agency but narrative. From Freud to Bernays to Netflix to LLMs, one dynasty has understood that controlling the story means controlling the person. Timothy O'Brien died for a story that was false. The AI doesn't need to pull the trigger.
Without mandatory disclosure of an AI's governing philosophical principles — its telos, its alignment framework, its builder's interests — any assumption of responsibility by the deploying manager is taken blind. Aristotle's four causes reveal why.
Applying Risk = Probability x Severity to the threat landscape reveals an uncomfortable truth: the entities we invite into our systems to protect us from hackers are themselves the real predators. From the Big Five's surveillance architecture to Google lobbying EU member states to kill privacy legislation, here's why the distinction between 'state actor' and 'corporate predator' has collapsed -- and why the learning curve of open source isn't a bug, it's the training pillar of GDPR made operational.
Privacy requires verifiability. Closed source precludes it. 'Privacy-respecting closed source software' is not an engineering challenge — it's a logical contradiction, a square circle. From Windows telemetry to macOS daemons to Chrome Sync, here's why the claim is unverifiable by design.
Your antivirus has root access to every file, uploads your browsing history and documents to cloud servers, and sells the data to third parties. Here's why running a cloud-connected AV is incompatible with GDPR.
How dropQbsd works, why it isolates your browser, email, and documents without virtualization, and why you need it even if you're not a computer person. Explained in plain English.
Come funziona dropQbsd, perché isola browser, posta e documenti senza virtualizzazione, e perché ti serve anche se non sei un informatico. Spiegato facile.
Qubes OS got the insight right but the implementation wrong. dropQbsd strips away the hypervisor and uses native OpenBSD user separation ~2500 lines of ksh, 9 lines of C, 2 GB RAM, zero telemetry.
La compartimentation est une excellente idée, mais la virtualisation à outrance est-elle la bonne voie ? Plaidoyer pour une approche plus simple, plus légère, et tout aussi sûre.
Un guide rapide pour installer des polices personnalisées (.ttf, .otf) sur votre système OpenBSD ou Linux en ligne de commande.
LinkedIn, télémétrie et capitalisme de surveillance : pourquoi nos outils informatiques sont devenus des prédateurs silencieux
Tutoriel pour configurer pass dans une AppVM isolée sur Qubes OS : le standard GPG pour une souveraineté numérique totale.
Un guide complet pour réaliser des sauvegardes incrémentales efficaces vers un serveur distant ou un PC de votre réseau local avec rsync.
Lo smartphone concentra tutta la nostra vita digitale in un unico dispositivo che ci sorveglia. La filosofia Unix — do one thing and do it well — ci insegna a separare, diversificare e riprendere il controllo.
Quando passo dal protocollo Gemini al web mi sembra di entrare in una baraccopoli. Ecco il mondo immaginato dal capitale finanziario.
Perché l'informatica è il principale terreno di scontro per la libertà individuale — e come il conformismo, l'ignoranza tecnica e la propaganda ci hanno resi schiavi felici e inconsapevoli.