OpenAI Finance Chief Warns Firms to Weigh Cybersecurity When Choosing Chinese AI
OpenAI CFO Sarah Friar advised companies to assess cybersecurity risks before adopting Chinese AI models, speaking at the International AI Summit in Dublin. Her comments came as lower-cost, open-weight models from China attract more business and developer interest. Friar suggested Western models may be preferable for security-sensitive uses, while noting her remarks did not prove all Chinese models are unsafe.
At the Dublin International AI Summit, OpenAI CFO Sarah Friar told Bloomberg Television that firms should weigh cybersecurity before choosing Chinese AI. Chinese developers now offer capable coding and analysis models at relatively low prices, and some open-weight versions can be downloaded for local use.
Anthropic's international head also cautioned at the same event that open models need careful safety review. OpenAI has delayed some releases for extra safety work. For companies, deployment choices, data access, hosting, permissions, and testing shape risk more than origin alone.
Businesses weighing cheaper Chinese AI may face harder procurement choices, especially those handling customer data, financial records, or proprietary code. Their employees and customers could be affected if security reviews are weak or inconsistent. The debate may also push Western and Chinese developers to compete more on safety assurances, while policymakers and industry groups could seek clearer deployment standards. Overall, the outcome may depend less on a model's origin than on how organizations govern access, hosting, and testing.