Google DeepMind finds no acceleration in AI research progress
Photo: Crypto Briefing
An internal audit tied to the Gemini 4 Argon safety report finds steady gains, heavy human review, and no sign of AI speeding up its own development Google DeepMind published its first comprehensive internal assessment of AI-driven machine learning research on October 14, 2026. The audit found no measurable acceleration in AI progress that can be credited to AI itself. The assessment appears inside the Gemini 4 Argon Safety & Security Report. Researcher Haydn Belfield led the work. The team tracked frontier Gemini capabilities with a metric it calls the General Capability Index, or GCI.
Source: Crypto Briefing
