Treasury Secretary says it's 'unacceptable' that Chinese open source AI models may have been built on 'IP theft'

Treasury Secretary Scott Bessent recently stated that the United States government is investigating whether Chinese open-source artificial intelligence models were built using stolen intellectual property from American firms. During a recent interview, Bessent characterized the practice of distilling US-based models to train foreign competitors as theft. While the administration maintains support for the open-source movement in principle, officials are now examining whether specific Chinese developers violated proprietary boundaries.

The investigation focuses on allegations that Chinese companies accessed frontier models from US firms to scrape data and research. This process, often referred to as distillation, allegedly allows overseas competitors to reach performance benchmarks equal to top-tier American models without the associated research and development costs. Bessent explicitly noted that authorities have identified watermarks of American models present within various Chinese offerings.

This position arrives as Chinese firm Moonshot AI prepares for the release of its Kimi K3 model. This technology has sparked debate among industry experts regarding competition and security. Some analysts argue that restricting access to open-source models could consolidate power among a small group of dominant US tech companies. Conversely, firms like Anthropic have argued that current practices allow for unauthorized training that circumvents standard licensing and terms of service agreements.

The Treasury Department maintains that if investigations confirm systematic intellectual property theft, sanctions remain a potential policy tool. Bessent emphasized that international participants must adhere to the same standards as domestic firms. As the government continues its review, the broader debate regarding the balance between open-source innovation and the protection of American technology assets continues to intensify.