Google's new forest map does more than identify trees. It translates satellite data into legal categories for supply-chain compliance. The useful product is not the prediction alone. It is the uncertainty, validation, history, and evidence around it.
Gemini 4 Argon ends Google's frontier-model silence with strong vendor benchmarks, serious internal-use claims, and a safety-first rollout. But trusted testers still stand between the announcement and public verification. Google has shipped evidence. It has not yet shipped confidence.
Six METR evaluators now have employee-like access inside Anthropic. That is a serious advance for external scrutiny, but it also exposes the missing half of the bank-examiner analogy: access produces evidence; public authority turns evidence into consequences.
Google DeepMind's Gemini Robotics 2 is not just a better humanoid demo. It is a reminder that physical AI needs whole-body control, progress tracking, local inference, hardware adaptation, and safety systems that work when the model can actually move things.