
The ARC-AGI-3 benchmark tested whether AI agents can generalize in unknown environments. Top models scored under 1% while humans solved 100%. The verdict: AGI is nowhere near here.

Google's latest quantum research paper names Algorand as a real-world example of post-quantum protocol deployment, sending the token up double digits in a single week.

Google DeepMind just published the most comprehensive map yet of how criminals and state actors are turning the open internet into a kill-zone for autonomous AI agents.

Google releases Gemma 4 в— a family of AI models under Apache 2.0. 31B model, MoE architecture, edge models on 1.5GB RAM.