Google is reportedly using Anthropic’s AI model, Claude, to evaluate and improve its own Gemini AI's performance. Contractors tasked with assessing Gemini’s outputs are comparing them against Claude under strict evaluation criteria. Internal communications revealed concerns over potential safety violations in Gemini's responses, especially regarding sensitive topics. Google has not confirmed whether it obtained Anthropic's permission for this comparative testing. A spokesperson stated that while model outputs are compared, Gemini is not being trained on Claude’s data. The race for superior AI models intensifies in the tech industry.