In a recent meeting in Singapore, Anthropic, a prominent AI research organization, declined a request from a Chinese think tank seeking access to its advanced AI model, Claude Mythos. This decision illustrates how the ongoing technological rivalry between the U.S. and China has intensified, especially in the field of artificial intelligence. Claude Mythos is categorized by Anthropic as having “digital-weapons-grade” capabilities, reflecting its potential impact on cybersecurity and defense strategies. For now, access to this model is limited to about 40 elite institutions, primarily situated in the U.S. and U.K., which focus on defense and intelligence operations.
This rejection represents a strategic choice by Anthropic. It has forfeited hundreds of millions in potential revenue by denying access to entities linked to the Chinese Communist Party, including firms under sanctions. The decision, made around late April or early May 2026, sheds light on the complex dynamics of global AI technology distribution.
In stark contrast, reports indicate that Chinese AI firms have engaged in aggressive attempts to breach Claude's defenses. These companies have executed large-scale cyberattacks, submitting over 16 million queries via fraudulent accounts, demonstrating a relentless pursuit of restricted capabilities. Notably, a group affiliated with the Chinese state has employed Claude's functionalities to conduct sophisticated cyber-espionage efforts targeting several organizations.
As the landscape of AI develops, monitoring such interactions between nations and within the technological community will be crucial for investors focusing on cybersecurity and the future of artificial intelligence technologies.