- OpenAI CFO Sarah Friar warned that businesses adopting low-cost Chinese AI models are taking on a cybersecurity risk, echoing concerns Anthropic has separately raised.
- Speaking at the International AI Summit in Dublin, Friar said companies should “think about the safety and security elements” before adopting Chinese models.
- Chinese AI labs now account for the largest share of generative model downloads on Hugging Face, driven by cheaper open-weight releases.
- Friar said OpenAI sees cybersecurity as a potential “big driver” of its own business growth.
What Happened
OpenAI Chief Financial Officer Sarah Friar warned that businesses choosing low-cost Chinese AI models are taking on a cybersecurity risk, speaking at the International AI Summit in Dublin, Bloomberg reported October 14, 2026. Friar said companies should “think about the safety and security elements” when weighing Chinese models and added: “we think you want to be on more Western models on the cyber front.”
Why It Matters
Friar’s warning echoes concerns Anthropic has separately raised about Chinese open-weight models and lands as both labs face mounting competitive pressure: Chinese AI developers now account for the largest share of generative model downloads on Hugging Face, driven by cheaper open-weight releases such as those tied to the DeepSeek-backed robot lab reported this month.
Technical Details
Friar argued the China pressure cuts both ways: US labs should not slow their own releases too sharply in response. OpenAI has held back some models to meet internal safety standards, but Friar said excessive restraint risks ceding control of the most capable AI tools to competitors operating under fewer such constraints.
Who’s Affected
Enterprises evaluating AI vendors face a harder tradeoff between lower-cost Chinese models and the Western-model security assurances Friar and Anthropic are both pushing. Friar separately identified cybersecurity services as a potential “big driver” of OpenAI’s own revenue growth going forward.
What’s Next
Watch whether OpenAI or Anthropic follow up with concrete enterprise cybersecurity offerings tied to this messaging, and whether other Western AI labs adopt Friar’s framing of Chinese models as a security liability in their own sales pitches.
