Nvidia Blackwell Powers Healthcare AI & Mistral Raises €3B & Chrome 149 Patches 28 Flaws

Nvidia's Blackwell infrastructure drives breakthroughs from healthcare AI to cheaper tokens, while Mistral secures €3B funding and Chrome fixes critical security gaps.

AI & ML

Nvidia and Abridge build domain-specific clinical foundation model on Blackwell infrastructure using de-identified data. The new model embeds clinical reasoning directly into foundation weights, enabling Northwestern Medicine's platform expansion to connect care delivery, payment, and evidence workflows. Read more
Nvidia Blackwell GPUs slash token costs 35x cheaper than older Hopper systems ($4.20 → $0.12 per million tokens), driving expected massive increases in token abundance and lower AI pricing across 2026. Read more
Mistral raises €3B at €20B valuation, nearly doubling its Series C valuation, positioning the European AI startup as a sovereign alternative to US-based OpenAI and Anthropic despite having raised only ~$4B to date. Read more
Deep Finance Analytics launches NEXT framework with 25 AI-native products for institutional finance, including PortIQ and Epsilon models designed for explainable, auditable risk analysis across portfolios. Read more

AI Coding Tools

GitHub updates npm v12 with three permissive defaults flipped to block install scripts, Git dependencies, and remote URLs by default, preventing malicious code execution during package installation. Developers can opt in via npm approve-scripts and prepare with npm 11.16.0+. Read more
DSIT pilots AI code remediation tools to modernize legacy systems, achieving nine months of work in four weeks in partnership with Defra, demonstrating dramatic acceleration potential across UK public sector. Read more
GitHub Copilot C++ modernization exits preview in Visual Studio 2026 v18.7, enabling developers to invoke @Modernize agent for modernizing C++ codebases. Read more

Languages & Runtimes

Swift TrueType hinting interpreter migration complete, moving Apple's font rendering subsystem from C to Swift for improved type safety and memory performance. Read more
Multi-agent orchestration patterns documented, with real-world lessons from evolving production chatbots into coordinator-based architectures and analyzing performance trade-offs. Read more

Browser & Web Platform

Chrome 149 patches 28 vulnerabilities, including 3 critical and 9 high-severity use-after-free memory safety bugs that could enable RCE or sandbox escape. Rolls out as 149.0.7827.114/.115 on Windows/macOS. Read more
Chrome releases experimental polyfills for built-in AI task APIs to support on unsupported platforms and browsers. Read more

Security

LangGraph critical vulnerability chain enables remote code execution via SQL injection in function handling, allowing attackers full server control. The open-source LangChain framework for multi-agent AI applications has patched three flaws. Read more
Over 400 Arch Linux AUR packages hijacked to deploy infostealer malware and eBPF rootkits, exploiting AUR's trust in package names and history over current maintainers. Read more