AI coding models

A developer known as drdator has released a web demo of City2026, a low-poly remake inspired by SimCity 2000, saying the game was built largely with the Claude Opus 5.5 AI model. The project is being shared on Hacker News, where readers are weighing both the retro city-building gameplay and what it says about AI-assisted game development.

Manifold Security published research asking whether large language models rely on moral reasoning when deciding if code is malicious. The finding: when asked to judge malware, models often reference ethical framing rather than purely technical analysis. The post is drawing attention among security researchers debating whether moral-flavored justifications make AI-assisted malware analysis more or less trustworthy.

A recorded talk by Linux kernel developer Greg Kroah-Hartman examines how large language models are affecting software security. Kroah-Hartman, who has long maintained kernel stable releases and driven the kernel's code-of-conduct and driver work, discusses the risks and implications of AI-generated code entering critical open-source infrastructure. The talk is drawing attention among developers weighing how LLM tooling should be handled in security-sensitive codebases.

A software development kit that brings AI-assisted coding to wireless microcontrollers has entered beta. The release adds an open-source community for developers and connects to enterprise MLOps tooling, aiming to ease development and deployment of machine learning models on connected embedded devices. It targets teams building wireless IoT products who want to move models from training to production.