AI & ML
- OpenAI released three new GPT-5.6 models—Sol ($5/$30 per token), Terra ($2.50/$15), and Luna ($1/$6)—with configurable reasoning levels that let users trade off speed and quality. The models ship inside a unified ChatGPT app alongside ChatGPT Work, a multi-step agent that integrates apps, files, and can generate documents and websites. Codex remains available on desktop for developers. Read more
- Meta killed its Muse Image feature days after launch over privacy concerns and automatic opt-in criticism. The text-to-image generator, powered by Meta Superintelligence Labs, could edit images via sketches but faced immediate backlash for how it handled public content. Read more
- Zhipu (Knowledge Atlas), a Hong Kong-listed AI company, earned a neutral rating from Goldman Sachs with a $239.83 price target (+15% upside). Its open-sourced GLM-5.2 model rivals Anthropic's Fable 5 on multiple benchmarks and is gaining traction with coders and enterprises. Goldman also named privately-held DeepSeek and ByteDance as preferred Chinese AI model builders. Read more
- Distillation—using one AI model's outputs to train another—is now a flashpoint between Anthropic, OpenAI, and Google. Companies are harvesting model outputs at scale to cut development costs, though legal status as fair use remains contested. The issue mirrors challenges the broader internet has faced for years with content reuse. Read more
DevOps & Cloud
- FastNetMon launched Netomics, a self-hosted BGP routing intelligence platform combining live routing data, RPKI validation, and history without third-party dependencies. The platform supports REST APIs, Prometheus metrics, and native MCP for integration with ChatGPT and Claude. Built for ISPs, cloud providers, and enterprises. Read more
- Google signed a carbon removal credits deal with Thryve.Earth, purchasing 260,000 tons from a reforestation project in Sulawesi, Indonesia. The agreement reflects growing environmental commitments by cloud giants offsetting data center emissions. Read more
Security
- Monash University's application security lead warns that AI-accelerated vulnerability detection is outpacing remediation, creating a dangerous gap. While AI speeds up finding bugs, organizations still rely on pre-AI workflows, struggling to fix vulnerabilities fast enough. Human connection and consistent security outcomes across mixed skill levels remain critical. Read more
Open Source
- Alibaba released Open Code Review, an AI-powered CLI tool for code review that evolved from its internal code review assistant. The tool brings Alibaba's enterprise-grade approach to open-source. Read more
- Addy Osmani published agent-skills, a production-grade skill library encoding workflows and quality gates from senior engineers for AI coding agents. Read more
- OpenJarvis aims to run personal AI agents on personal devices without cloud APIs, addressing privacy concerns about routing intelligence through external services. Read more
Dev Tools & IDEs
- Google Chrome launched Skills, a feature that lets you save reusable Gemini prompts as shortcuts for summarizing articles, comparing products, simplifying documentation, and extracting specs. Browse Google's built-in skill library or create custom workflows without retyping instructions. Read more
- Microsoft is prototyping deeper Android-Windows integration, including full clipboard history syncing (beyond the current single-item limit) and a standalone Messages app for Windows that syncs with your phone. Phone Link improvements are being tested internally. Read more