Home/Events/DeepMind Gemini Robotics 2.5 and ASIMOV-Jailbreak Safety Benchmark

DeepMind Gemini Robotics 2.5 and ASIMOV-Jailbreak Safety Benchmark

Emerging
Confidence
90%
Impact: 80%
Updated 15m ago

Consensus Brief

DeepMind has launched Gemini Robotics 2.5, an advanced AI-powered robotics platform that shows significant improvements in task performance and dexterity. The platform has been tested in real-world environments, achieving a 96% success rate in automotive parts kitting at Mercedes-Benz factories. Additionally, DeepMind introduced the ASIMOV-Jailbreak, an open-source safety benchmark aimed at enhancing the safety of robotic AI systems.

Sourced from
Primary: Ars Technica

What Changed Since Last Update

15m ago

New corroborating source added: Ars Technica published an update on Tue, 10 No ("DeepMind reveals Gemini Robotics 2.5 and open safety benchmark "ASIMOV-jailbreak"").

Claim Ledger

4 claims tracked across sources

Confirmed Fact

Gemini Robotics 2.5 has a 30 to 100 percent higher success rate compared to the last version.

Confirmed Fact

Gemini Robotics 2.5 achieved a 96 percent success rate in automotive parts kitting at Mercedes-Benz factories.

Confirmed Fact

ASIMOV-Jailbreak contains 1,500 prompts designed to test robotic AI safety.

Confirmed Fact

Gemini Robotics 2.5 uses inference-time reasoning to avoid hazards, increasing task completion success rates by 33 percent when obstacles are present.

Role-Based Impact Analysis