OpenAI CFO Sarah Friar Warns Companies Over Chinese AI Cyber Risks
OpenAI CFO Sarah Friar has warned businesses about potential cybersecurity risks from low-cost Chinese AI platforms, urging them to use Western-developed systems for critical workflows. She also revealed that OpenAI is delaying its Astra model's release while assessing possible autonomous cyber capabilities, highlighting growing concerns over AI safety and enterprise security.
OpenAI Chief Financial Officer Sarah Friar has issued a warning to global enterprises regarding the digital security risks associated with adopting low-cost Chinese artificial intelligence platforms. Speaking at the International AI Summit in Dublin, Friar urged commercial organisations to rely on Western-developed infrastructure for critical operational workflows rather than budget alternatives.
The executive framed the choice of underlying technology as a fundamental cybersecurity consideration. As per a report by The Times of India, Friar stated that companies utilizing low-cost Chinese systems expose themselves to serious digital vulnerabilities, making Western platforms a far safer alternative for mission-critical tasks. OpenAI To Double Workforce to 8,000 Employees by Late 2026: Report.
OpenAI Holds Back Astra Model Amid Security Concerns
During her presentation, Friar also addressed broader industry debates regarding the pace of artificial intelligence development. She cautioned against artificial delays in innovation, arguing that excessively slowing down American technology firms could create a strategic imbalance over who controls powerful systems.
However, Friar revealed that OpenAI is currently holding back the release of its internal model named Astra. The decision stems from safety evaluation protocols, as company researchers cannot yet rule out that the system possesses critical autonomous cyber capabilities that could lead to real-world operational disruptions.
Industry Competitions and Open Source Vulnerabilities
The security warnings arrive as open-weight models developed in China capture the largest share of generative model downloads on platforms such as Hugging Face. Representatives from rival laboratory Anthropic supported Friar's position at the summit, arguing that open-source model distributions inherently carry elevated operational risks for enterprise deployment. OpenAI Will Not Go Public in 2026; CEO Sam Altman Cites AI Safety Focus.
The debate highlights complex industry dynamics, particularly after open-source platform Hugging Face deployed GLM 5.2, a Chinese open-weight model, to conduct forensic evaluations following a security incident earlier this year. As global adoption accelerates, technology leaders continue to weigh the cost benefits of budget models against potential cybersecurity exposures.
(The above story first appeared on LatestLY on Oct 15, 2026 07:32 AM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).
