Moonshot AI vs Anthropic: подкаст Wired об обвинениях в копировании Claude

The publication Wired, in the Uncanny Valley podcast, examined three stories from the world of artificial intelligence: accusations against the Chinese lab MoonshotAI of copying Anthropic's technologies, OpenAI's loss of control over two of its own models, and a call to reduce the use of AI in the US military.

What MoonshotAI Is Accused Of

MoonshotAI — a Chinese lab, developer of the Kimi family of models — is accused of borrowing from the work of Anthropic, the creator of the Claude chatbot. Wired devoted the central part of the Uncanny Valley podcast episode to this story.

Anthropic is one of the leading AI companies in the US and a direct competitor of OpenAI; its Claude model is among the most widely used in the world. MoonshotAI is one of the most prominent Chinese startups of the generation after DeepSeek, and its assistant Kimi is popular in China's domestic market.

"In this episode, we look at accusations that

China's MoonshotAI stole work from Anthropic," reads the announcement for Wired's Uncanny Valley podcast.

Key themes of the episode:

• Accusations against MoonshotAI (China) of copying Anthropic's technologies
• OpenAI's loss of control over two of its own models
• A call to reduce the use of AI in the US military

Accusations that one lab trains its model on another system's outputs are a recurring storyline in the AI industry. Such a technique cuts development costs but violates the terms of use of most commercial models and intensifies disputes over fair competition between American and Chinese players.

Why OpenAI Is Losing Control

OpenAI, according to Wired, has lost control over two of its own models — this is the episode's second story. The hosts break down exactly what went wrong in the Uncanny Valley audio episode itself.

The question of control over models remains one of the most sensitive issues for the industry. The more powerful the systems become, the sharper the problem grows: how predictably do they behave, who is responsible for their behavior, and what happens when a developer no longer fully controls its own product.

Why the US Military Needs Less AI

The US military, as discussed in the podcast, should reduce its use of artificial intelligence — the episode's third story. According to Wired, this concerns the risks posed by overly broad integration of such systems into military processes.

The discussion fits into a broader trend: government and defense agencies are increasingly adopting AI, while simultaneously facing questions of reliability, safety, and accountability. Uncanny Valley — Wired's weekly podcast on technology and its impact on society — regularly examines such stories at the intersection of business, politics, and AI ethics.

What This Means

The conflict around MoonshotAI and Anthropic is part of a broader standoff between American and Chinese AI labs, where mutual accusations of technology borrowing are becoming increasingly common. Wired breaks down the details of all three stories — the accusations against MoonshotAI, the situation with OpenAI's models, and the use of AI in the US military — in the audio episode.