Anthropic rejects Chinese think tank access to Claude Mythos

Techub News reports, citing CryptoBriefing, that Anthropic has stated it rejected a request from a Chinese think tank to access its "digital weapon-grade" AI model Claude Mythos. The model is currently only available to approximately 40 defense and intelligence agencies in the U.S. and the U.K., and the company claims it has thus foregone hundreds of millions of dollars in revenue.

The rejection occurred during a meeting in Singapore in late April or early May. Additionally, it is reported that a Chinese AI company launched a large-scale attack on Claude, sending over 16 million queries through fraudulent accounts to gain access to restricted features; a certain China-linked state organization is also alleged to have used the model to conduct AI-orchestrated cyber espionage activities.