Cette semaine, je sépare le contenu en trois sections. Et, je crois, les suivantes suivront le même modèle. La première, pour des articles que je n’attrape pas à travers mes mécanismes de vigie pour le podcast. La seconde, les articles que j’attrape à travers ma vigie pour le podcast, mais qui n’ont pas été retenu, considérant qu’ils n’ont pas une composante qui m’intéresse d’aborder sur mon podcast. Finalement, la 3e, les articles qui ont été retenus pour le podcast et où, si vous le désirez, pouvez écouter mon opinion sur le sujet des articles.
Section 0 : Rattrapage
- A quote from Tom Dale
- How StrongDM’s AI team build serious software without even looking at the code
- AI Doesn’t Reduce Work—It Intensifies It
- GLM-5: From Vibe Coding to Agentic Engineering
- Apple bets big that AI models will become commodities; may offer hot-swappable Siri
- A quote from Andrew Deck for Niemen Lab
- How Generative and Agentic AI Shift Concern from Technical Debt to Cognitive Debt
- Gemini 3 Deep Think
- A quote from Thoughtworks
Section 1 : Hors podcast
- Do Markets Believe in Transformative AI? - Marginal REVOLUTION
- AI fatigue is real and nobody talks about it
- GPT-5 bests human judges in legal smack down • The Register
- This dev made a Llama with three inference engines • The Register
- Two different tricks for fast LLM inference
- Breaking the Spell of Vibe Coding – fast.ai
- Chatbots Make Terrible Doctors, New Study Finds
- Stop generating, start thinking - localghost
- AI chatbots don’t improve medical advice, study finds • The Register
- GitHub - antirez/voxtral.c: Pure C inference of Mistral Voxtral Realtime 4B speech to text model
- https://www.bicameral-ai.com/blog/tech-debt-meeting
- [2512.20798] A Benchmark for Evaluating Outcome-Driven Constraint Violations in Autonomous AI Agents
- The Singularity will Occur on a Tuesday - Cam Pedersen
- GLM-5: From Vibe Coding to Agentic Engineering
- Claude ouvre ses outils pro à tous (et y’a zéro pub) - Korben
- Anthropic’s AI‑built C compiler is not all that impressive • The Register
- https://assets.bwbx.io/documents/users/iqjWHBFdfxIU/rRmql_jJcxb4/v0
- Why I’m not worried about AI job loss - David Oks
- AI agents turned Super Bowl viewers into one high-IQ team — now imagine this in the enterprise
- AI Agent Lands PRs in Major OSS Projects, Targets Maintainer…
- [2602.12125] Learning beyond Teacher: Generalized On-Policy Distillation with Reward Extrapolation
- Fragments: February 13
- Niche Hunt — App ideas people actually want
Section 2 : Podcast
- AI threat modeling must include supply chains, agents, and human risk
- OpenClaw instances open to the internet present ripe targets
- Microsoft boffins show LLM safety can be trained away
- Augustus - Open-source LLM Vulnerability Scanner With 210+ Attacks Across 28 LLM Providers
- AI-Generated Text and the Detection Arms Race
- AI agents can spill secrets via malicious link previews
- Claude add-on turns Google Calendar into malware courier
- The First Signs of Burnout Are Coming From the People Who Embrace AI the Most
- Claude and OpenAI fight over ads while Google monetizes
- Prompt Injection Via Road Signs
- NanoClaw solves one of OpenClaw’s biggest security issues — and it’s already powering the creator’s biz
- Microsoft: Poison AI buttons and links may betray your trust
- Anthropic safety researcher quits, warning ‘world is in peril’
- Cyber Model Arena
- AI bot seemingly shames developer for rejected pull request
- AI Weaponization: State Hackers Using Google Gemini for Espionage and Malware Generation
- Misconfigured AI could shut down a G20 nation, says Gartner
- AI Agents ‘Swarm,’ Security Complexity Follows Suit
- OpenAI has deleted the word ‘safely’ from its mission – and its new structure is a test for whether AI serves society or shareholders
- Pentagon used Anthropic’s Claude during Maduro raid
- How AI could eat itself: Using LLMs to distill rivals
- Your Friends Might Be Sharing Your Number With ChatGPT