OpenAI, Anthropic, and Google have begun sharing sensitive security intelligence through the Frontier Model Forum, forming an informal defense alliance against model cloning. The move follows Anthropic identifying over 16 million fraudulent Claude exchanges linked to approximately 24,000 accounts traced back to competing companies — including DeepSeek, Moonshot AI, and MiniMax. The intelligence-sharing arrangement represents a notable shift: companies that fiercely compete on capabilities are now cooperating on protecting their intellectual property. The Frontier Model Forum, originally created to coordinate AI safety research, is expanding its mandate into commercial IP defense.