Latest Articles from the world of Artificial Intelligence
27 July 2026
Jacobian lens: what Claude 'thinks' and does not say
When an artificial intelligence paper talks about "mind," "workspace," and "unspoken thoughts," we are facing useful metaphors to make complex concepts accessible, but also potentially misleading for the very public they should simplify things for. On July 6, Anthropic published a new study that uses exactly this kind of language, and in the following days the web was divided between those who read the news as the antechamber of a conscious machine and those who dismissed it as engineering disguised as philosophy. The truth, as often happens, lies in the middle, but to find it one must carefully separate the data from the words chosen to describe them.
24 July 2026
Loop Engineering: current trend or the next leap?
In mid-June, a post by Peter Steinberger, a developer known for building and then selling PSPDFKit, circled among insiders for weeks: AI agents are no longer prompted, the loops that run them are designed. A few lines, a tone of revelation, and within a few days the term "loop engineering" appeared everywhere, from Analytics Vidhya to corporate blogs specialized in agent infrastructure like Requesty. Even Boris Cherny, among the architects of Claude Code, upped the ante by talking about a leap comparable to that from compiled source code to autonomous agents, as BDTechTalks recounts while reconstructing the genesis of the debate.
22 July 2026
Why AI agent benchmarks underestimate real capabilities
For years we have treated an AI agent's capability as a fixed number, almost like a sprinter's time: you run, you time it, you write the result on a board. But what if that number depended on the breath we allow the runner before stopping them? A report published on July 2, 2026, by the AI Security Institute, the UK government's AI safety institute, suggests that this is exactly the hidden error in many agent evaluations: we are timing them with a whistle that blows too early.
20 July 2026
Colibrì: the mini engine that runs a 744B model on consumer hardware
Every week, the local open-source artificial intelligence community seems to look for its favorite conversation topic, and the one that just passed seems to have elected a GitHub repository with a curious name and an almost absurd promise: running GLM-5.2, a 744-billion-parameter Mixture-of-Experts model, on a computer with just 25 GB of RAM. The project is called Colibrì and is the work of a single developer, JustVugg, who in the acknowledgments section of his README admits with disarming honesty that he wrote and tested everything on a twelve-core laptop. No lab, no cluster, no hardware sponsorship. Just a technical obsession carried to its conclusion.
17 July 2026
DSpark: DeepSeek's bet on speed that doesn't betray quality
DeepSeek has not only presented a new approach to speculative decoding; with DeepSpec, it is trying to transform it into a reproducible industrial pipeline. The paper is called DSpark, the acronym is yet another in a long series that the Chinese lab churns out almost quarterly, and the temptation to scroll through it distractedly is strong. That would be a mistake, because behind the acronym hides a very concrete question: how much can the inference of a language model really accelerate if the model that generates the draft responses stops being naive, and if the system that checks them learns not to waste time on those destined for the bin.
15 July 2026
I taught AI to stand guard: how I built a zero-cost security system
In the TV series 'Person of Interest', a superintelligence nicknamed simply "the Machine" monitors every corner of the planet through cameras, microphones, and sensors of all kinds, identifying threats before they materialize. It's science fiction, of course. But the underlying idea—using computer vision to understand what's happening in an environment—is now accessible to anyone with a PC and a webcam. Less than 250 lines of code. No subscription, no cloud, no video traveling the world. Just a notification on your phone, with a photo, when someone enters the house.
13 July 2026
AI is running, the world is walking: what the first UN scientific report says
There is a scene in Serial Experiments Lain in which the protagonist discovers that the network to which she has always been connected has stopped being a tool and is becoming an environment, something that includes and defines her without anyone having ever consciously decided it should happen. Reading the first independent United Nations scientific report on artificial intelligence brings to mind exactly that feeling, a system that has expanded faster than our ability to describe it, and now someone is finally trying to do so methodically.
10 July 2026
Ornith-1.0 35B local: the unknown beating everyone
There is a moment, in every session with a new model downloaded locally, when you understand if you are facing a toy or a work tool. With Ornith-1.0-35B that moment arrived at the second prompt, when I uploaded a blurred photo of a company Excel sheet expecting the usual vague answer, and I found myself with a real financial statement analysis, complete with liquidity warning signals. From then on, the test session took a different turn than usual.
08 July 2026
GLM-5.2: the Chinese open-weight closing the gap, at least in coding
When Z.ai, the laboratory everyone in China still knows as Zhipu AI, released GLM-5 last February, the message was already clear: the GLM series wasn't just aiming to be competitive, it was aiming to be relevant for those who build software. That 744-billion-parameter model with a Mixture-of-Experts architecture had put performance on the table that didn't look out of place next to the big proprietary ones, with the far-from-secondary difference of open weights under the MIT license. Just over four months later, on June 13, 2026, Z.ai raised the stakes again: GLM-5.2 arrived first on the GLM Coding Plan tiers, the subscription plans dedicated to developers, and then, on June 16, on public API and on Hugging Face with freely downloadable weights.
06 July 2026
Do you really need the latest AI model? Or are you just upgrading your tech ego?
Do you really need the latest state-of-the-art model to do your daily work? If your answer is "yes," are you sure you haven't been misled by big tech marketing? Anthropic has just launched Claude Opus 4.8, OpenAI already has the next version of ChatGPT in the works, and everyone is pushing us to chase the most recent model as if our productivity depended on that last decimal of a benchmark. Yet, for 90% of daily activities, an efficient model like, for example, DeepSeek V4 Flash costs a fraction of the price and does exactly the same thing. Does this remind you of something? It does to me, and I will try to tell you about it.
03 July 2026
Headroom: the token compressor that doesn't use AI to compress AI
There is a silent paradox at the heart of many modern artificial intelligence systems. Engineers build sophisticated agents capable of reasoning, planning, and coordinating complex sequences of actions. Then they look at the monthly API bill and realize that the biggest expense is not reasoning, not creativity, not even accuracy. It is traffic. The sheer, banal volume of text that components exchange with each other.
01 July 2026
Last30days: when a code agent becomes a social search engine
Before writing anything about AI, I opened eleven browser tabs: Reddit, X, YouTube, Hacker News, GitHub, and a few industry newsletters. Two hours later, after three coffees, I had found three truly useful posts. The rest was noise: blog articles optimized for search engines, opinions of people paid to have them, classic galleries of the "ten AI tools that will change your life" type, written with the same depth as an assembly instruction sheet for IKEA furniture.
29 June 2026
KV cache and the weight of attention: three paths, one problem
Llama-3.1-70B, running in BF16 precision, accumulates about 0.31 megabytes of KV cache for every single token processed. With a context of 128,000 tokens, the bill rises to 40 gigabytes, an already uncomfortable figure. With one million tokens—the standard toward which the most recent models are pushing—it exceeds 300 gigabytes: more than the 140 gigabytes occupied by the model weights themselves. It is a detail that overturns a widespread intuition, that the critical memory in a large language model is that of its parameters. It is no longer so, or at least not always. The memory that truly scares those who design long-context inference systems is that of the KV cache, the archive where the model keeps in store the key and value representations of every token already seen, so as not to have to recalculate everything from scratch for every new word generated.
26 June 2026
Courses cancelled in favor of AI: China rewrites the university
There is a scene, recounted by the South China Morning Post, that is worth more than a thousand statistics: a recent industrial design graduate explains that his course was suspended because artificial intelligence hit that sector hard, where modeling and rendering can now be carried out, in large part or entirely, by an algorithm. It is not an isolated case but the symptom of a transformation that has swept through the entire Chinese university system in a handful of years and with a determination that has no equivalent in the West.
24 June 2026
The algorithmic war. Fable 5 is the finger, Europe is the Moon
On June 12, 2026, the United States Department of Commerce sent Anthropic a letter that, within the industry's halls, circulated with the speed of news that truly inspires fear: mandatory suspension of the Fable 5 and Mythos models for every foreign citizen, inside or outside American borders. The official justification is national security. Groups linked to Beijing reportedly gained access to Mythos, the version without guardrails designed for cybersecurity applications, by bypassing access control systems. A serious breach, if confirmed, without a doubt. But stopping at this news, as many are doing, obsessed with the names Fable 5 and Mythos as if they were characters in a dystopian series, is like looking at the finger and not the moon.
22 June 2026
BenzUp: I created an app without writing a single line of code
There is a precise moment when an idea stops being a bar fantasy and becomes something concrete. In my case, that moment had as protagonists, in order: the cost of gasoline, an American engineer annoyed by an expensive beer in Dublin, and an artificial intelligence model accessible to anyone with an internet connection. The result is called BenzUp, it is free, it does nothing revolutionary, and perhaps that is exactly why it is worth telling.