Chinese company Qihoo 360 has made bold claims regarding the development of an AI system called “Tolongfeng.”
CEO Zhou Honyi said the technology is a “Chinese version” of Anthropic’s powerful Mythos AI, with characteristics of both a “cyber nuclear weapon” and a strategic deterrent for the digital age.
This system is reportedly able to discover software vulnerabilities. Zhou claims that Tulongfeng has already identified more than 3,000 vulnerabilities, some of which have been classified as “high risk.”
Zhou also pointed out that while the technology may not match Mythos in raw power, when combined with Qihoo’s existing technology, it could create unprecedented hacking capabilities comparable to Anthropic’s systems.
He called the development of AI-driven cyber systems a new nuclear arms race and said, “Historically, nuclear weapons have served as the primary strategic deterrent, but the future of national security will likely be defined by vulnerability discovery capabilities.”
Qihoo 360 is currently on the US blacklist due to suspected ties to the Chinese military and intelligence services. However, the company denied all these accusations.
The claim came after the Five Eye security alliance, made up of the US, UK, New Zealand, Canada and Australia, issued a stark warning that adversaries were just months away from using AI to launch “overwhelming cyberattacks”.
Alan Woodward, a cybersecurity expert at the University of Surrey, made it clear that China would develop such a system.
“It might not be as good as Mythos, but it shows that these things happen anyway,” he said.
