What happened
A rising San Francisco human resources startup, Rippling, has been using a blacklisted Chinese AI lab's model, setting off alarm bells due to security concerns. The model's maker, Z.ai, was added to a US government entity list months earlier over its ties to the Chinese military.
Rippling, which services over 20,000 US companies, has been testing Z.ai's GLM 5.2 model, which is 85% cheaper than other AI models but offers nearly identical performance. However, experts warn that using Chinese AI models poses significant security risks, including deliberate data exfiltration and intelligence collection.
Other US companies, including giants like DoorDash and small startups, are also adopting Chinese AI models due to their low cost and high performance. However, this trend has raised concerns among experts, who warn that the potential risks are significant and could be difficult to detect.
