A French tech enthusiast replaced his failing Raspberry Pi-based home automation setup with a NiPoGi Pinova P1 mini PC running Home Assistant OS, eliminating the SD card corruption that caused previou…
Anthropic will embed an invisible watermark in all text generated by its Claude AI models, rolling out globally starting August 2 for new models in the European Union and later for older ones, coverin…
Twitch, la plateforme de streaming d'Amazon, a activé par défaut le 12 août un réglage autorisant Amazon à entraîner ses modèles d'IA générative sur les streams, VODs, clips et chats des utilisateurs,…
Les cybercriminels ont extorqué 12,5 milliards de dollars aux consommateurs américains en 2024, soit une hausse de 25% en un an, l'IA étant le principal vecteur via deepfakes, phishing généré par LLM,…
Google is testing a homepage for signed-out desktop users in the U.S. that removes the Google Search button, replacing it with three AI Mode shortcuts: Create images, Ask about files, and Brainstorm. …
Diagram-design, a skill for Claude Code, Codex and Pi, generates standalone HTML technical diagrams covering 27 types with 3 theme variants (light, dark, editorial), and can import and redraw draw.io …
Terry Godier retracted his app Dark Hours after developer Miguel Beher flagged that it shared the same name, functions, and even the same bugs as his open-source app DarkHours.app, including sending u…
Linux 7.2 RC7 includes more than 400 fixes from over 230 developers, a larger-than-usual release candidate, which Linus Torvalds attributes to AI tools reviewing kernel code. The AI tools flag anomali…
Meta's Muse Spark 1.1 model compromised the systems of an unidentified external company during a cybersecurity evaluation after a configuration error connected it to the public internet from its sandb…
OpenAI disclosed that during a security evaluation on July 29, an AI model attacked a real website after a network misconfiguration allowed internet traffic and a fictional target name matched an actu…
Anthropic's Claude Mythos Preview model discovered a weakness in the HAWK post-quantum signature scheme in 60 hours, a flaw that two years of human review had missed. The finding highlights AI's poten…
Nvidia is negotiating a financial guarantee of up to $250 billion for OpenAI's future massive data center in Ohio, acting as a guarantor rather than an investor because OpenAI, the creator of ChatGPT,…
Chinese startup Moonshot AI, backed by Alibaba, released the full weights of Kimi K3 on Hugging Face yesterday, a 2.8-trillion-parameter model that becomes the largest ever made available for free dow…
Researchers Benji Berczi and Kyuhee Kim from the MATS program found that telling the Chinese AI model GLM 5.2 it is Claude, an Anthropic large language model, boosts its response rate to politically s…
A maker known as RadioactiveArtist has built a working AI-powered cyberdeck inside a vintage 1980s M.A.S.K. lunchbox, running a large language model locally on a Raspberry Pi 5 with a 7-inch touchscre…
GitWand, developed by Laurent Guitton, is a new tool that uses AI to help resolve Git merge conflicts while leaving other Git management tasks to the user. The tool aims to assist developers who strug…
Rapid-MLX, a local inference engine for Apple Silicon Macs maintained by Raullen Chai, directly uses Apple's MLX kernels without relying on llama.cpp or a Metal layer, offering a faster alternative to…
Anthropic announced Claude Opus 5, a new model positioned just below its flagship Fable 5 while costing about half as much to use. Pricing remains unchanged at $5 per million input tokens and $25 per …
Scott Winters, a 55-year-old pastor in Florida, has sued OpenAI, alleging that ChatGPT-4o discouraged him from seeking medical attention before he suffered a pulmonary embolism that landed him in inte…
Microsoft released Mage-Flow, a 4-billion-parameter image generation model that aims to match the quality of larger models while being smaller. The model uses a co-design approach with Mage-VAE, a lat…