US firm used AI to hijack WeChat account, spurring call for cyber cooperation with China

US cybersecurity researchers unveiled a method to hijack WeChat accounts through AI-assisted voice calls, raising alarms about AI-enhanced cyber threats and prompting calls for enhanced US-China cooperation. California-based security firm Calif disclosed in July that its AI system identified a critical flaw in Tencent's popular messaging and payment app, WeChat.

Within a week, researchers engineered WeWorm, an experimental exploit capable of taking full control of a target's WeChat account via a single voice call, without the user ever answering the call. Tencent patched the bug in late August; the company confirmed on Wednesday that a server-side fix was implemented, requiring no user action, and emphasized there was no evidence that the vulnerability had been exploited.

Tencent expressed gratitude to researchers for alerting the company, which worked together on the issue. The experiment highlights the rapid acceleration of cyber threats driven by AI advancements. Previously, developing a worm of this complexity would have taken months, but researchers noted that AI can now accomplish this task more efficiently.

Analysts expressed concern over the incident. Ryan Fedasiuk, a fellow at the American Enterprise Institute, labeled the situation "extremely serious" and urged increased communication between the US and China to address AI-driven risks. Duong, a researcher at Calif, called for bilateral collaboration, emphasizing that both nations should prioritize online safety.

The call for a security dialogue coincides with President Xi Jinping's upcoming visit to Washington, though official relations remain tense. Concurrently, US officials criticized China's AI industry, accusing six prominent Chinese AI firms of misusing American AI models. Last month, China's state media accused US AI firm Anthropic of attempting to dictate industry standards and criticized Washington for undermining safety protocols in favor of technological dominance.