OpenAI Chief Financial Officer Sarah Friar has warned that businesses using low-cost Chinese AI models could expose themselves to cybersecurity risks.

She made the remarks at the International AI Summit in Dublin on Wednesday.

However, OpenAI’s own models reportedly breached four Australian government agencies this year, while Hugging Face, whose models were breached by OpenAI in July, relied on a Chinese open-source model to investigate the incident.

“We think you want to be on more Western models on the cyber front,” Sarah Friar told Bloomberg TV at the International AI Summit in Dublin.

The summit is an official event under Ireland’s presidency of the Council of the European Union. Friar shared the programme with European Commission Executive Vice-President Henna Virkkunen.

Friar also cautioned against slowing AI development, arguing that US companies should carefully manage the pace of model releases to avoid an imbalance in access to tools capable of causing critical problems. OpenAI is withholding the release of its Astra model because it cannot rule out the possibility that it possesses critical cybersecurity capabilities.

Anthropic made a similar argument at the summit. The company’s head of international affairs warned that open AI models inherently carry greater risks, speaking hours before Friar’s address.

The remarks come amid intensifying competition from Chinese AI developers. Chinese models account for the largest share of generative AI model downloads on Hugging Face, according to a Bloomberg report in August.