JJ Jasser argues that Kimi K3, a Chinese AI model, misleadingly markets itself as ‘open-source’ despite lacking the transparency required by established open-source definitions. The piece contends that genuine open-source AI requires public disclosure of architecture, training code, model weights, and training data, criteria that K3 fails to meet. Jasser warns that without access to training data, users cannot understand why models omit or distort information, making such models unsuitable for education regardless of benchmark performance, and characterizes the mislabeling as ‘open-washing’ that serves China’s geopolitical positioning as an AI leader.
