Anthropic is lying: Moonshot is not routing to Claude (twitter.com)

Anthropic has come under scrutiny for its claims regarding the routing of user requests through its AI models, specifically alleging that the Chinese market's use of AI gateways—dubbed "transit stations"—is centered around their services such as Claude and Opus. However, it's argued that these claims are misleading. Users in China are forced to navigate barriers to access these models, often resorting to foreign phone numbers for account creation, leading to frequent bans. Transit stations circumvent these obstacles by aggregating subscriptions and exposing them as APIs, enabling users to access models at significantly reduced prices. This situation has significant implications for the AI/ML community, particularly regarding ethical data practices and market integrity. Transit stations not only provide ultra-cheap model access but also exploit vulnerabilities in payment systems and significant promotions, making revenue through pooled subscriptions and potentially selling user interaction data for training purposes. These practices raise concerns about transparency and the quality of AI outputs, as users might be unknowingly interacting with lower-tier models while being promised access to more advanced systems like Opus or Claude. This controversy underscores the complexities and ethical considerations in the burgeoning AI marketplace, especially in regions with stricter access and regulatory challenges.