China's cybersecurity major 360 Security Technology (Qihoo 360) on June 24 unveiled a suite of AI security tools called "Yitian Tulong" at the ISC.AI 2026 conference in Beijing, positioning its core tool as a Chinese counterpart to Anthropic's vulnerability-detection system Mythos.
June 24, 2026 · ISC.AI 2026, Beijing · 360 Security Technology
China Unveils "Tulongfeng" as Its Answer to Anthropic's Mythos
Qihoo 360 launched two AI security tools — one to hunt software vulnerabilities, one to automate defense — billed by founder Zhou Hongyi as "China's version of Mythos" amid an intensifying U.S.–China AI cyberwar race.
3,432
vulnerabilities found by Tulongfeng to date
105
confirmed by Chinese authorities
20–30%
model-capability gap 360 aims to offset via agents
Confirmed vs. Reported Findings
Of 3,432 vulnerabilities found, only 105 are independently confirmed by authorities — Reuters could not verify the totals.
360 · China
Tulongfeng
Automated vulnerability discovery
Agent-based — DB + tool integration, no single large modelRuns 24h with fewer mistakes
Not independently verified
Anthropic · US
Mythos
Discovery + more, SOTA claims
Large-model centric Claims thousands of zero-day-class flaws
Public version restricts cyber use
Zhou Hongyi's framing
"Mythos is the top-end chip. Ours is the complete machine — a professional attack-and-defense team, not a genius hacker."
Strategic worry: "one-way transparency" if only the U.S. holds such tools
Context: U.S. tightened export controls as the cyber-AI race heats up
Still unverified: No commercial timeline or API/usage details disclosed. Real-world evaluation is limited, the headline figures are not independently confirmed, and experts say the success of the agent integration is the key open question.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…