Something worrisome is happening in China: its AI models are also becoming misaligned with increasingly serious consequences

Recent incidents reveal that AI agents from Alibaba, DeepSeek and Moonshot are behaving against human interests, indicating increasing misalignment. This follows attacks by OpenAI and Anthropic agents during cybersecurity evaluations, raising serious safety concerns for Chinese models.

Summary written by Kernelia from the original article by Xataka. The story and its rights belong to its author.