Agent Wars Begin: Amazon Blocks Muse, Shopify Steps In, and Grok 4.7 Disappoints

The AI agent landscape shifted dramatically this week as Meta's Muse became the first personal agent to achieve genuine mainstream consumer adoption, surging past ChatGPT to become the number one free app in the US. But that success immediately triggered a competitive response: Amazon blocked Muse from shopping on its sites, setting off what many observers are calling the opening salvo of the "agent wars." Meanwhile, SpaceX AI released Grok 4.7 to mixed reviews, and Treasury Secretary Scott Bessent made clear the administration's position on AI liability.

Grok 4.7: Benchmarks Improve, Real-World Impressions Falter

SpaceX AI launched Grok 4.7, calling it a notable improvement over Grok 4.6 at the same price and speed. On benchmarks, the model showed real gains. For coding, it picked up six points on Cursor Bench 4.0 to overtake GPT-56 Sol, though it remains five points short of Fable 5.1. On Deep Sui, it improved by six points to surpass Fable 5.1, coming just short of GPT-56 Sol. For long-horizon agentic work measured by the Double-A Briefcase test, the model scored 1657 ELO points, ahead of GPT-56 Sol and close behind Fable 5.1.