The Rise of AI-Driven Cybersecurity in China: A Response to Anthropic's Mythos

On April 7, 2026, US start-up Anthropic unveiled its advanced artificial intelligence model, Claude Mythos Preview, which has garnered significant attention for its ability to autonomously identify and exploit cybersecurity vulnerabilities. This model has not been publicly released; instead, it is accessible to a select group of organizations through Project Glasswing, which includes major US companies like Cisco and JPMorgan Chase. While Anthropic has labeled China as an "adversarial nation," its announcement has nonetheless sparked considerable interest within China's cybersecurity sector.

South China Morning PostSouth China Morning PostBloombergChinese Cybersecurity Firm’s AI Hacking Claims Draw Comparisons to Claude Mythos

In the wake of Mythos' announcement, shares of prominent Chinese cybersecurity firms, including Qi An Xin, Sangfor Technologies, and 360 Security Technology, experienced a notable rise, reflecting heightened expectations for AI-driven cybersecurity solutions. Austin Zhao, a senior research manager at IDC China, noted that the market's reaction underscores the sector's keen interest in Mythos, with many firms anticipating a surge in demand for advanced cybersecurity measures. James Gong, legal director at Bird & Bird, indicated that the technology could lead to increased cybersecurity expenditures as firms invest in personnel and infrastructure.

South China Morning PostBloombergSouth China Morning PostChinese Cybersecurity Firm’s AI Hacking Claims Draw Comparisons to Claude Mythos

360 Security Technology has claimed to develop an AI-powered "Vulnerability Discovery Agent" that has identified nearly 1,000 previously unknown vulnerabilities, including critical flaws in widely used software such as Microsoft Office. This claim positions 360 as a competitor to Anthropic's Mythos. The company's AI system reportedly played a crucial role in its success at the Tianfu Cup, a major Chinese hacking competition, where it identified a significant number of vulnerabilities. However, while these capabilities are impressive, Eugenio Benincasa, a cybersecurity researcher at ETH Zurich, cautioned that 360's AI does not yet match the reasoning capabilities attributed to Mythos.

Chinese Cybersecurity Firm’s AI Hacking Claims Draw Comparisons to Claude MythosBloombergSouth China Morning PostSouth China Morning Post

China's cybersecurity landscape is shaped by legislation requiring private companies to report vulnerabilities to government agencies before public disclosure. This regulatory framework channels elite security research into state intelligence, potentially enhancing the operational impact of AI capabilities developed by firms like 360. Benincasa highlighted that this integration between private firms and state actors could facilitate the direct translation of cybersecurity improvements into offensive cyber capabilities.

Bloomberg

Despite the advancements, some experts express skepticism regarding the authenticity of 360's claims, particularly concerning the attribution of certain vulnerabilities. Additionally, while Chinese models are rapidly improving, they are still perceived to lag behind their US counterparts in overall cyber capabilities. The competitive landscape raises questions about the ethical implications of AI in cybersecurity, especially given the potential for state-sponsored cyber intrusions.

South China Morning Post

As AI technology continues to evolve, the competition between US and Chinese cybersecurity firms is likely to intensify. The developments surrounding Anthropic's Mythos and China's response through companies like 360 underscore the growing significance of AI in cybersecurity. The implications of these advancements extend beyond corporate competition, potentially reshaping the global cybersecurity landscape in the years to come.

South China Morning PostSouth China Morning PostBloombergChinese Cybersecurity Firm’s AI Hacking Claims Draw Comparisons to Claude Mythos