Anthropic’s newest public model skips China.
On June 9 the company launched Claude Fable 5, a toned‑down version of its larger Mythos system. The model is deliberately engineered to refuse API calls from any AI lab located in China. Anthropic announced the restriction as a compliance move, but the loudest objections came from engineers and researchers inside the firm, who say the gatekeeping limits the model’s usefulness and complicates collaborations.
The decision matters because it signals a deeper split in the AI community over geopolitical access controls. By blocking Chinese research groups, Anthropic may cede market share to rivals that adopt a more open stance, while also inviting scrutiny over how private firms enforce national‑security policies.
The episode shows how quickly technical choices can turn into internal PR battles, especially when they intersect with geopolitics.