256 Newsroom — Uganda's Digital News Infrastructure
Business

Treasury Secretary says its 'unacceptable' that Chinese open source AI models may have been built on 'IP theft' - Business Insider

Share
Treasury Secretary says its 'unacceptable' that Chinese open source AI models may have been built on 'IP theft' - Business Insider
Image · Business Insider

What the report says

Business Insider reported that Treasury Secretary Scott Bessent said Tuesday that the Trump administration is examining whether Chinese open-source artificial intelligence models were developed using intellectual property taken from U.S. competitors. In a Fox Business interview cited by Business Insider, Bessent said the administration supports open-source AI generally, but views the alleged use of U.S. model outputs or research to train rival systems as unacceptable and potentially subject to sanctions.

The comments add a federal enforcement angle to a dispute already playing out inside the AI industry. Business Insider noted that Anthropic has accused Chinese companies, including Moonshot AI, MiniMax and DeepSeek, of using a technique known as distillation to draw on Claude at scale while avoiding detection and violating its terms. Bessent did not identify specific firms, but said U.S. officials were finding signs, described as watermarks, of American large language models in Chinese systems.

The issue has become more prominent as Chinese open-weight models gain attention from developers and investors. Business Insider pointed to Moonshot AI’s Kimi K3, which has drawn interest because benchmark results suggest strong coding and general performance compared with leading proprietary models from U.S. companies. Moonshot has said it plans to release full weights for Kimi K3 on July 27, while users can currently access the model through its website.

The debate matters because open-source or open-weight models can be downloaded, modified and deployed widely, potentially shaping competition in AI. Business Insider also noted disagreement among U.S. tech figures over whether Washington should discourage use of Chinese models, while some critics question complaints from AI companies whose own training practices have faced copyright challenges.

Read the full report at Business Insider →

Loading debate for this article…

Other publishers covering this story

No additional verified coverage is currently clustered with this report.

Related reporting