Chinese AI developers released 16 models in one month, led by DeepSeek and Xiaomi, while Anthropic CEO Dario Amodei called on frontier AI companies to slow development. Nikkei Asia reported the releases on Wednesday, amid Amodei’s warning that AI capabilities are advancing faster than safeguards. A DeepSeek engineer who worked on the company’s V4.1 models rejected the call, saying he did not trust Anthropic or OpenAI to keep advanced AI open and affordable.

The disagreement also has a commercial dimension: DeepSeek and other Chinese firms have built followings with open-weight models, which developers can download, customise and run themselves instead of paying to use closed systems. The South China Morning Post reported that Chinese observers questioned the motives behind calls to slow AI development. Bloomberg reported this week that Moonshot AI closed a funding round at a valuation of about $50 billion, while DeepSeek was close to completing a $12 billion raise.

In February, Anthropic accused DeepSeek, Moonshot AI and MiniMax of running “distillation” campaigns. It said they used about 24,000 fraudulently created accounts to generate more than 16 million exchanges with its Claude model. Distillation means training a model on another system’s outputs. OpenAI and US officials have made similar accusations against DeepSeek and Moonshot.

Security concerns have also featured in calls to slow development. Forbes reported last month that a Chinese-speaking hacker used AI agents, including DeepSeek and Kimi models and an older Claude version, to attack as many as 100 organisations. Anthropic and OpenAI have cited AI risks, including a breach at Hugging Face, in urging a slowdown.

Chinese AI developers released 16 models in one month, led by DeepSeek and Xiaomi, while Anthropic CEO Dario Amodei called on frontier AI companies to slow development. Nikkei Asia reported the releases on Wednesday, amid Amodei’s warning that AI capabilities are advancing faster than safeguards. A DeepSeek engineer who worked on the company’s V4.1 models rejected the call, saying he did not trust Anthropic or OpenAI to keep advanced AI open and affordable.

The disagreement also has a commercial dimension: DeepSeek and other Chinese firms have built followings with open-weight models, which developers can download, customise and run themselves instead of paying to use closed systems. The South China Morning Post reported that Chinese observers questioned the motives behind calls to slow AI development. Bloomberg reported this week that Moonshot AI closed a funding round at a valuation of about $50 billion, while DeepSeek was close to completing a $12 billion raise.

In February, Anthropic accused DeepSeek, Moonshot AI and MiniMax of running “distillation” campaigns. It said they used about 24,000 fraudulently created accounts to generate more than 16 million exchanges with its Claude model. Distillation means training a model on another system’s outputs. OpenAI and US officials have made similar accusations against DeepSeek and Moonshot.

Security concerns have also featured in calls to slow development. Forbes reported last month that a Chinese-speaking hacker used AI agents, including DeepSeek and Kimi models and an older Claude version, to attack as many as 100 organisations. Anthropic and OpenAI have cited AI risks, including a breach at Hugging Face, in urging a slowdown.