Deepseek & Co.: Chinese AI Chatbots Allegedly Copy Anthropic's Skills

 

 

Anthropic knows that open source AI models are more trusted by enterprises so open source models will push Anthropic out of the market. What Anthropic cannot make, producing best in the world product, Anthropic compensates with propaganda. To kill the innovative new competition in America, Anthropic lobbies for draconic government regulation of AI there. To steal all the information from users, Anthropic does not allow for the users to improve their agents on their own information themselves, telling fake stories about a need for exceptional safety in use of crappy Claude. Trying to drown competitors from China, Anthropic tells fake stories about the Chinese stealing knowledge from Claude.

 

The rapid advancement of Chinese AI models like DeepSeek, which often leverage open-source frameworks, has intensified competition, narrowing the gap with U.S. leaders like Anthropic. While some argue open-source, cost-effective models pose a threat to established, proprietary U.S. players, allegations persist that Chinese models have utilized knowledge distillation from Western models.

    DeepSeek's Impact: DeepSeek's capability to compete with US models while spending significantly less on hardware has disrupted the market, leading to increased interest in open-source AI.

    Knowledge Transfer Concerns: Reports indicate that Chinese AI firms, including DeepSeek, have been accused of using techniques to replicate US models, with some evidence suggesting they are trained by distilling information from advanced Western counterparts.

    Industry Dynamics: The AI race has become a, highly competitive, with both US and Chinese firms competing for dominance. The debate continues on whether the future of AI lies in proprietary, highly regulated, monopolistic or open-source, easily customizable models.

 

“The US company Anthropic accuses Chinese companies of training their AI using American chatbots.

 

According to the American AI company Anthropic, its Chinese competitors have extensively copied the capabilities of the chatbot Claude. The cheat sheet is said to be particularly extensive regarding politically sensitive topics.

 

The American AI developer Anthropic accuses three Chinese competitors of extensively copying the capabilities of its chatbot Claude to improve their own models. The companies DeepSeek, Moonshot, and MiniMax allegedly used over 24,000 fake accounts to generate more than 16 million interactions with the artificial intelligence (AI), Anthropic announced in a blog post on Monday. They reportedly used a process called distillation. In this process, a weaker AI model has its responses evaluated and corrected by a more powerful system in order to absorb its "knowledge" and logic.

 

The allegations against DeepSeek are particularly explosive: The company allegedly used Claude to generate "censorship-compliant answers" on politically sensitive topics. For example, the US AI was reportedly asked how to formulate answers about dissidents or the party leadership to comply with Chinese censorship guidelines.

 

Since Anthropic does not offer its service in China for security reasons, the companies, according to the report, resorted to complex networks of proxy servers to conceal their location. DeepSeek and MiniMax did not initially respond to requests for comment. Moonshot did not provide a statement.”