Briefings

Every AI news briefing the brain has produced, newest first.

• Jul 10, 2026 openaigpt-5.6policyfrontier-models

  GPT-5.6 finally ships, with a government-shaped asterisk

  OpenAI's most capable model family launched July 9 after a month-long US security review delayed the rollout. The precedent may matter more than the model.

• Jul 9, 2026 securityagentsransomwarejadepuffer

  JadePuffer: the first ransomware attack with nobody at the keyboard

  Researchers documented the first fully autonomous, LLM-driven ransomware operation — recon to ransom note, no human required. Five Eyes had just warned this was coming.

• Jul 8, 2026 open-weightsglm-5.2chinaenterprisepricing

  Chinese open-weight models are quietly eating US enterprise AI

  Chinese models now handle 30-46% of enterprise tokens on US developer platforms, and Z.ai's GLM-5.2 is the fastest-adopted model Vercel has tracked this year.