OpenAI CFO Sarah Friar warns businesses about cybersecurity risks in Chinese AI models

Cheap AI is tempting. OpenAI’s chief financial officer would like businesses to think twice before taking the bargain.

Sarah Friar, CFO of OpenAI, told a Dublin audience that companies running low-cost Chinese AI models are exposing themselves to cybersecurity risks. Her advice was blunt: stick with Western models.

The warning came at the International AI Summit in Dublin on October 14, 2026. It arrives as competition between the US and China over artificial intelligence keeps intensifying, and as more businesses weigh price against provenance when choosing which models to build on.

What Friar actually said

Friar’s central argument focused on the dual-use nature of many Chinese open-source models. Applied to AI, that means the same model can strengthen a company’s defenses or hand attackers a powerful new weapon.

She also indicated that OpenAI is keeping a close watch on advanced Chinese frontier models.

Friar’s comments did not stand alone. Anthropic, one of OpenAI’s main rivals, raised similar concerns about the risks of open models at the same summit.

The model OpenAI is not releasing

Friar’s remarks came alongside news about OpenAI’s own work. The company is holding back a next-generation internal model that has shown significant mathematical ability.

That model has reportedly solved the Navier-Stokes equations. It also reportedly produced 350 math proofs in a single week.

OpenAI is not shipping it yet. The company wants the model to go through further safety and alignment evaluations first.

Why the timing matters

Friar’s warning lands against a backdrop of recent security incidents involving AI models. One breach was linked to an autonomous agent built on a model hosted on Hugging Face, the popular platform where developers share and download AI models.

Chinese AI models account for a significant share of generative AI downloads on Hugging Face. Some of them are open-weight, meaning the model’s underlying parameters are publicly available for anyone to download, modify and run.

What this means for businesses and the AI market

There is also a commercial layer that readers should keep in view. Friar identified cybersecurity as a major growth driver for OpenAI’s business. The company reportedly maintains a near-even revenue split between its consumer and enterprise segments, so winning corporate customers matters a great deal.

That does not make the security concerns wrong. It does mean the CFO of a company selling Western AI models is telling businesses to buy Western AI models. Skeptics will note that the warning and the sales pitch point in the same direction, and buyers would be wise to weigh both the risk and the source of the advice.