July 13, 2025 1:36 pm CEST Elon Musk’s xAI launched its record-breaking Grok 4 model this week, but the celebration is overshadowed by doubts. The powerful new AI arrived amid controversy, including a formal apology for its predecessor’s “horrific” antisemitic meltdown and the discovery that Grok 4 consults Musk’s personal views on sensitive topics. The rollout, which also saw the resignation of X CEO Linda Yaccarino, comes as xAI reportedly seeks a staggering $200 billion valuation. This series of events puts a sharp focus on the company’s high-stakes strategy: pursuing frontier AI performance while grappling with severe ethical and safety crises. Grok 4: A New Benchmark King with a Contentious Crown On July 10, through a livestream on X, xAI officially unveiled Grok 4 and its more powerful “Heavy” variant, positioning them as the world’s most capable AI models. The launch, was backed by a slate of impressive, independently verified results showing a significant leap in reasoning and problem-solving abilities. The new model demonstrated its dominance on several key benchmarks. It nearly doubled the previous record on the difficult ARC-AGI-2 reasoning benchmark, a test designed to measure human-like abstract intelligence. Grok 4 achieved 15.9% accuracy, a major advance in a field where most models struggle to reach single digits. It also scored a perfect 100% on the American Invitational Mathematics Examination (AIME).