Breakthroughs in Chinese AI models threaten U.S. lead in tech race - PBS

What the report says

PBS NewsHour reported on July 22 that recent advances by Chinese artificial intelligence companies are intensifying concern in Washington and Silicon Valley over the United States’ position in the AI race. In an interview with Geoff Bennett, Wall Street Journal tech reporter Amrith Ramkumar discussed new models from Moonshot AI and Alibaba that are described as open-weight, free to use and comparatively low-cost while nearing the performance of leading U.S. systems from companies such as OpenAI and Anthropic.

According to the PBS segment, Ramkumar said Chinese developers have narrowed the gap despite having less access to advanced computing resources, in part by finding engineering workarounds and using techniques such as distillation, where newer models learn from outputs or behavior of other systems. He said the cost advantage could make Chinese models attractive to U.S. companies seeking to reduce AI spending, even as some executives warn about security and competitive risks.

The report also highlighted policy questions facing the White House. Ramkumar said administration officials are weighing possible responses, including trade restrictions or blacklists for some Chinese firms, while parts of the tech industry argue the U.S. should compete by developing its own open models rather than limiting access. He noted that some American businesses are already using Chinese AI systems, which could complicate any government action.

PBS also reported that the White House is planning a broad shift in federal research and development funding, placing AI more centrally across agencies and prioritizing individual researchers over large universities. Ramkumar said the approach could significantly affect colleges that depend on federal research funds, while officials maintain that human oversight would remain part of the process.