China's Z.ai Claims It Can Match Anthropic's Mythos 5 on Cybersecurity Capabilities

The AI race between the United States and China just shifted into a new gear. In a development that sent shockwaves through Silicon Valley and Washington alike, Chinese AI company Z.ai has publicly claimed that its cybersecurity-focused AI model now matches the capabilities of Anthropic's highly classified Mythos 5 system — one of the most powerful AI models ever built for security applications.

This isn't a fringe claim from a little-known startup. The story has been picked up by both the Wall Street Journal and the New York Times, lending it a degree of credibility that demands serious attention from anyone following the global AI competition.

The Claim: Z.ai Matches Anthropic's Mythos 5

Z.ai, a Chinese artificial intelligence company specializing in cybersecurity applications, has announced that its latest AI model achieves parity with Anthropic's Mythos 5 across a range of cybersecurity benchmarks. According to the company's published findings, their model excels in:

• Vulnerability detection: Identifying zero-day exploits and software vulnerabilities at speeds comparable to Mythos 5
• Threat intelligence analysis: Processing and correlating massive threat datasets to identify emerging attack patterns
• Automated incident response: Generating real-time defense strategies and countermeasures against active cyberattacks
• Network security monitoring: Continuous surveillance and anomaly detection across complex enterprise networks
• Malware reverse engineering: Deconstructing and analyzing sophisticated malware samples with deep behavioral understanding

The implications of this claim are enormous. If verified, it would mean that the United States no longer holds an insurmountable lead in one of the most strategically critical domains of AI — cybersecurity. This is a field where AI capabilities directly translate to national security advantages.

What WSJ and NYT Are Reporting

The Wall Street Journal ran the headline: "China Has Matched Anthropic in Cybersecurity, Resetting AI Race" — a striking statement from one of America's most influential business publications. The article details how Z.ai's model was tested against Mythos 5 on standardized cybersecurity benchmarks and achieved results that analysts described as "within the margin of error" of Anthropic's system.

The New York Times followed with its own reporting, noting that "Chinese A.I. Models Close the Gap With Anthropic and OpenAI" — framing the development as part of a broader trend rather than an isolated incident. The NYT piece highlighted that this isn't just about Z.ai; it reflects a systemic acceleration in China's AI development capabilities across multiple domains.

Both outlets noted the timing is particularly significant. Anthropic's Mythos 5 was designed to be the gold standard in AI-powered cybersecurity — a system so advanced that the US government took the extraordinary step of pulling it offline and imposing export restrictions. The fact that a Chinese company is now claiming parity with a system that Washington deemed too dangerous to export speaks volumes about the pace of Chinese AI development.

360's Parallel Claim: Another Chinese Challenger

Z.ai isn't alone in this race. China's 360 Security Technology — one of the country's largest cybersecurity firms — has separately announced that it has developed AI tools that match the capabilities of Anthropic's Mythos system. This parallel development from a major, publicly traded Chinese company adds significant weight to the broader narrative.

360's claim is particularly noteworthy because the company has deep ties to the Chinese government and military-industrial complex. Their cybersecurity AI capabilities are not theoretical — they are actively deployed across Chinese enterprises, government agencies, and critical infrastructure. When 360 says it has matched Mythos-level capabilities, it's not just a lab result; it's a statement about operational readiness.

The convergence of claims from both Z.ai and 360 suggests that China's cybersecurity AI ecosystem is maturing rapidly. This isn't a one-company achievement — it's an industry-wide advancement that reflects massive state investment, abundant data resources, and a strategic national priority to achieve AI self-sufficiency.

Mythos 5 Pulled Offline: The Export Ban Factor

Understanding why this development matters requires context about Anthropic's Mythos 5 itself. The system was originally developed as a next-generation cybersecurity AI capable of proactive threat detection, autonomous response coordination, and deep analysis of nation-state-level cyber threats.

However, the US government determined that Mythos 5's capabilities were too sensitive for public deployment or international distribution. The system was pulled offline and subjected to strict export controls. The rationale was straightforward: an AI this powerful in the hands of adversaries could fundamentally compromise US cybersecurity infrastructure.

The irony is sharp. By restricting Mythos 5, the US inadvertently created a vacuum — a gap that Chinese companies have now rushed to fill. While Anthropic's system sits behind government restrictions, Chinese alternatives are free to develop, iterate, and deploy without the same constraints. The export ban, designed to protect American advantages, may have inadvertently accelerated Chinese AI development by removing competitive pressure and giving Chinese companies clear runway to catch up.

What This Means for the US-China AI Race

The US-China AI race has entered a phase that can only be described as competitive parity in cybersecurity. This has profound implications for both nations and the global technology landscape.

For the United States, the loss of a clear lead in cybersecurity AI represents a strategic vulnerability. Cybersecurity is not just a commercial market — it's a matter of national defense. If Chinese AI can match American AI in defending against (and potentially conducting) cyber operations, the traditional American advantage in digital warfare erodes significantly.

For China, this development validates years of massive investment in AI research. Despite US export controls on advanced chips, restrictions on AI research collaboration, and efforts to limit Chinese access to cutting-edge hardware, Chinese companies have found ways to develop world-class AI capabilities. The Z.ai and 360 claims suggest that hardware limitations may be less of a constraint than Washington assumed.

For the rest of the world, the message is clear: the AI monopoly is over. Organizations and governments can no longer assume that the most capable AI systems are exclusively American. The cybersecurity tools available in the global market will increasingly include Chinese alternatives that compete at the highest level.

Cybersecurity AI: Why It Matters

Cybersecurity is arguably the most consequential application of AI in 2026. Here's why:

• Scale of threats: The volume of cyberattacks has grown exponentially. Human security analysts simply cannot process the volume of alerts, logs, and threat data generated by modern enterprise networks. AI is not optional — it's essential.
• Speed of attacks: Modern cyberattacks unfold in milliseconds. AI-powered defense is the only way to detect and respond in real-time at machine speed.
• Nation-state actors: Both the US and China are investing heavily in offensive and defensive cyber capabilities. AI gives nation-state actors the ability to automate attacks, discover vulnerabilities at scale, and adapt defenses dynamically.
• Critical infrastructure: Power grids, financial systems, healthcare networks, and transportation systems all depend on cybersecurity. AI-powered protection of these systems is a matter of public safety.

When a Chinese company claims to match the most advanced American cybersecurity AI, it's not just a technology story — it's a national security story with global implications.

The Bigger Picture for AI Development

Z.ai's claim about matching Mythos 5 fits into a larger pattern that the AI industry has been watching closely. Over the past 18 months, Chinese AI models have made dramatic advances across multiple benchmarks. What was once a clear American lead — exemplified by OpenAI's GPT models and Anthropic's Claude series — has narrowed considerably.

Several factors are driving this convergence:

• Massive state investment: The Chinese government has poured billions into AI research through national programs, provincial initiatives, and state-backed venture capital
• Talent pipeline: China produces more AI researchers and engineers than any other country, creating a deep talent pool
• Data advantages: Chinese companies have access to enormous datasets from China's 1.4 billion population, providing training data at a scale that American companies cannot easily match
• Strategic focus: Cybersecurity AI is a national priority for China, driving focused development in this specific domain
• Silicon workarounds: Despite chip restrictions, Chinese companies have developed creative solutions including model optimization, inference efficiency improvements, and custom hardware alternatives

The cybersecurity dimension of this race is particularly telling because it demonstrates that China isn't just catching up in general-purpose AI — it's achieving parity in specialized, high-stakes domains where the implications extend far beyond commercial applications.