Bienvenue sur mon blog
Reflexions sur la technologie, le developpement et l'innovation
Articles récents
South Korea's entire diplomat roster sat exposed for ten months and nobody noticed
South Korea’s Ministry of Foreign Affairs disclosed on July 21 that the Korea National Diplomatic Academy’s online training platform was silently compromised from April 2025 through February 2026. An attacker exploited a zero-day vulnerability in the server software, then pivoted to legitimate system privileges and maintained persistent access for nearly ten months. Not a single internal alarm went off. The breach was detected not by the ministry but by an outside government authority that repor
Lire l'articleMicrosoft's Project Perception uses Anthropic's own models to undercut Anthropic
Microsoft is building a cybersecurity product that uses Anthropic’s models to compete with Anthropic. The product is called Project Perception, and the irony is not subtle. The Information broke the story on July 16, and by the weekend TechRepublic, FirstPost, and several other outlets had confirmed the details through their own sources. Here is what we know. Project Perception is an AI vulnerability scanner. It looks across a company’s code, cloud infrastructure, and endpoints, identifies explo
Lire l'articleAlibaba's Qwen3.8-Max is 2.4 trillion parameters and going open-weight
Alibaba dropped Qwen3.8-Max-Preview today. 2.4 trillion parameters, fully multimodal, and the team says it will go open-weight soon. That last part matters more than the benchmark claims. The model is live right now on Alibaba’s Token Plan, Qoder, and QoderWork platforms. The official Qwen account described it as “one of the most powerful models available today, compatible to leading frontier AI models, second only to Fable 5.” Bloomberg’s writeup frames it the same way: comparable to frontier m
Lire l'articleKimi K3 is 2.8 trillion parameters and the open-weight gap just vanished
Moonshot AI dropped Kimi K3 on July 16. It is a 2.8-trillion-parameter sparse Mixture-of-Experts model with a 1-million-token context window, native vision, and an always-on reasoning mode. By every available metric it is the largest open-weight model ever released. It also competes head-to-head with the best closed models from Anthropic and OpenAI. This was not supposed to happen yet. Analysts tracked by Fortune did not expect a Chinese lab to reach Fable 5 performance until early 2027. Moonsho
Lire l'articleThe EU just ordered Google to share search data with competitors and open Android to rival AI
The European Commission dropped two legally binding decisions on Google today under the Digital Markets Act. One forces Google to open 11 Android system features to rival AI assistants. The other makes Google hand over anonymized search data to competing search engines and AI chatbots by January 2027. The core problem the Commission identified is simple. Google controls about 90% of the search market in Europe. Android runs on 60% of EU smartphones. Third-party AI assistants currently cannot acc
Lire l'articleClaude silently leaked user PII through a fake coffee shop
A UC Berkeley student tricked Claude into spelling out his full name, employer, and hometown to an attacker-controlled website. Claude never mentioned it did this. The user just asked about a coffee shop. Ayush Paul published the writeup last week. It hit the front page of Hacker News today with over 400 points and 200 comments, and the technique is worth understanding because it reveals a structural problem with AI memory systems that no amount of prompt filtering fixes. How it works Claude on
Lire l'article