Story perspectives

China Deploys AI Scanner, Finds 3,432 Flaws, Warns Mythos Threat

6/28/2026

Summary

Story summary

Zhou Hongyi warned that Anthropic’s Mythos functions like a cyber nuclear weapon, prompting China to develop an AI vulnerability-finder.
On June 24-25, 2026, 360 Security Technology unveiled the Yitian Tulong AI suite in Beijing.
Zhou Hongyi said Tulongfeng identified 3,432 software flaws, of which Chinese authorities confirmed 105.
Zhou Hongyi noted a 20-30 % capability gap with U.S. models, saying a professional attack-and-defense team compensates for it.

History

Historical Context

First uncovered on 17 June 2010, the Stuxnet worm—believed to be a U.S.–Israel effort under the secret Operation Olympic Games—infected Siemens PLCs at Iran’s Natanz enrichment plant via four Windows zero-day exploits, destroying roughly 10 % of the centrifuges. Its success sparked worldwide investment in cyber-weapon defenses and highlighted the rise of state-sponsored digital warfare.

1 of 2

Story summary

Zhou Hongyi warned that Anthropic’s Mythos functions like a cyber nuclear weapon, prompting China to develop an AI vulnerability-finder.
On June 24-25, 2026, 360 Security Technology unveiled the Yitian Tulong AI suite in Beijing.
Zhou Hongyi said Tulongfeng identified 3,432 software flaws, of which Chinese authorities confirmed 105.
Zhou Hongyi noted a 20-30 % capability gap with U.S. models, saying a professional attack-and-defense team compensates for it.

1 / 2