this week in ai, openai is turning chatgpt into a super app with coding and agents, while apple prepares a major siri overhaul using google gemini. the trump administration discusses taking an equity stake in openai, and sriram krishnan steps down as white house ai advisor. on the research front, a new study questions ai consciousness, and a time series method improves forecasts without retraining.

  1. openai plans chatgpt super app with coding and agents - this move signals openai's push to compete with anthropic and boost paid subscriptions by bundling more tools.
  2. siri gets ai overhaul at wwdc 2026 - apple's reliance on google gemini for siri's revamp highlights the growing partnerships in ai, along with new agent and visual tools.
  3. trump administration eyes equity stake in openai - this unusual proposal could create a public wealth fund, sharing ai profits with citizens and reshaping government-industry ties.
  4. sriram krishnan leaves white house ai advisor role - krishnan's departure after 18 months marks a shift in the administration's ai policy leadership.
  5. bees and chatgpt: the consciousness question - this research challenges how we assess consciousness in ai and animals, with implications for ethics and future machine design.
  6. gated inference-time context optimization for time series models - gitco offers a practical way to boost forecast accuracy without costly retraining, useful for real-world time series applications.

other notable stories include google's new ai tools for thrift shopping, openai's lockdown mode against prompt injection, and the ladybird browser project halting public pull requests over ai-generated code concerns. a hackathon project explored multi-model finance drama, while python file writing basics and openai codex voucher issues rounded out the week.