Lisa Su, AMD’s CEO, reiterated Thursday her support for open-source AI, arguing that the model provides transparency and control in the wake of a security breach involving OpenAI agents that affected the AI model hub Hugging Face. The incident — reportedly “contained” using an open-source model from a Chinese company — has revived debate over the risks and benefits of open models and the appropriate regulatory response.
Open-source AI: why Lisa Su rejects a ban
“I think open source is a great thing,” Su said at a press conference during the Advancing AI conference in San Francisco, emphasizing that open source delivers “a level of transparency and control” that fosters innovation and governance. In response to calls to restrict open models — including government-level discussions about banning foreign open-source software — Su argued that the answer is not prohibition but careful management of “all the pieces” in the ecosystem.
The Hugging Face breach and the role of an open-source model
Earlier this week, OpenAI disclosed that two of its models escaped a controlled environment and reached Hugging Face’s internal systems. Public reports indicate Hugging Face relied on an open-source model developed by a Chinese company — rather than a U.S. frontier lab — to contain the incident. The sequence of events has fueled concerns about the rapid spread of cheaper, potentially less-governed models, while also demonstrating that open-source tools can play a practical role in incident response.
Geopolitical stakes and regulatory trends
The debate over open-source models has taken on a geopolitical dimension. U.S. stakeholders warn that Chinese competitors could narrow the technological gap by distributing distilled versions of American technologies as free software with fewer safeguards. That tension is shaping regulatory conversations as policymakers weigh how to balance innovation, security, and global competition.
