AI writes the code, but who checks the safety?

Detailed summary

Large models are accelerating code generation, yet ensuring its security and correctness remains a challenge. During the 2026 World Internet Conference Wuzhen Summit, Zhao Lu,…

Large models are accelerating code generation, yet ensuring its security and correctness remains a challenge. During the 2026 World Internet Conference Wuzhen Summit, Zhao Lu, chief scientist at Nineye, said open-source software security once relied on manual audits, traditional static analysis and dynamic testing. However, as large models become increasingly involved in coding, securing open-source software is emerging as a new challenge for the industry.

Opening of the article, excerpted from China Daily. To read the rest, open the article on the publisher's site.