Anthropic dominates the day's AI news with a $65B Series H that pushes its valuation toward $1 trillion, alongside the release of Claude Opus 4.8 and new 'dynamic workflows' for Claude Code that let agents restructure their own task plans mid-run. Cognition's Devin is reportedly raising $1B at a $26B valuation, MiniMax teased a sparse-attention M3 model promising huge long-context speedups, and Microsoft is said to be building its own coding model to compete again. Apple is reportedly trying to squeeze a distilled version of Google's Gemini onto iPhone hardware to power a revamped Siri, while ByteDance joins the custom-silicon trend to escape chip supply constraints. On the security side, researchers documented an attacker using an LLM agent to pivot from a CVE to an internal database in just four steps, plus fresh Iran-linked campaigns and a four-year-old Gitea bug exposing private container images. For engineers, the through-line is that AI capability, capital, and attack surface are all scaling together, and yesterday's productivity metrics for both builders and defenders are getting stale fast.
Frontier AI: Money and Models
Big Tech's Infrastructure Moves
Engineering Leadership & Productivity
Security: AI Agents Cut Both Ways