The Trump administration has imposed new restrictions on the release of private AI models developed by companies like Anthropic and OpenAI, exercising a kill-switch over models controlled by single companies and based on proprietary data. This move has intensified calls for open-source AI development in the United States.

Supporters of open-source models argue that these alternatives, which rely on publicly accessible data, could help the U.S. maintain a competitive edge. They warn that the White House’s unprecedented intervention in frontier AI labs might inadvertently benefit China, which offers more affordable open-source models globally.

AI experts highlight the urgency of expanding open-source efforts domestically to prevent China from capitalizing on a shortage of American frontier AI models. Maele noted the pressure companies face, referencing an incident last month when the Trump administration gave Anthropic only 90 minutes to withdraw its Claude Mythos 5 and Fable 5 models following cybersecurity concerns raised by Amazon.

Anthropic and CEO Dario Amodei have historically opposed open-weight models, cautioning during a 2023 Capitol Hill visit that the growth of open-source AI is "going down a dangerous path." They argue that private models allow for post-deployment modifications, including revoking user access.

Meanwhile, industry voices express frustration with the current environment. On CNBC’s "Squawk Box," Karp said, “Something has gone completely wrong,” describing a widespread enterprise sentiment of wasting time on tokens without gaining value, while fearing intellectual property theft.

Vinny Troia, CEO of intelligence firm Shadow Nexus, criticized the administration’s approach as "price gouging" that incentivizes a shift toward open models. He told The Hill, “What’s happening with the White House is only going to further compound the problem.”

Research from venture capital firm Andreessen Horowitz indicates that 80 percent of developers worldwide using open-source tools are building with Chinese models, underscoring concerns about U.S. leadership in AI development.

Sources