Anthropic, the US AI safety company behind the Claude model family, is racing OpenAI and Google to define frontier AI while navigating export controls, model IP disputes, and a near-US$1 trillion valuation ahead of a prospective IPO.