Dear Reader,
Anthropic’s latest report alleges that seven Chinese AI labs used fraudulent accounts, proxy networks, and other techniques to extract capabilities from its Claude models.
Alibaba reportedly generated over 151 million exchanges to train Qwen, while Moonshot AI and DeepSeek allegedly routed millions of user queries to Claude, exposing sensitive government, military and corporate data to US servers.
Meanwhile, OpenAI has temporarily paused new subscriptions to its $200-a-month Pro plan after demand for its newly-launched Astra AI model exceeded the company’s expectations.
The surge in usage has reportedly put pressure on OpenAI’s computing capacity, prompting it to limit new Pro sign-ups while it works to accommodate demand.
