Gemini Robotics 2 Brings Google's AI Into the Physical World

Google DeepMind launched Gemini Robotics 2, an AI system for controlling robots in the physical world.

It combines vision-language and vision-language-action models to help robots perceive, reason, and carry out tasks like object manipulation and shelf tidying.

The company showed autonomous behavior across multiple robot types and said it trained the system with teleoperation, videos, and simulations.

DeepMind also introduced ASIMOV-Agentic, a new benchmark focused on robot safety.

The release signals Google’s push toward broadly capable physical AI, while underscoring the risks of giving frontier models control of real machines.