For three months, Anthropic's Claude Code was quietly checking whether the developer using it was in China or working through a Chinese AI lab, then hiding the answer inside the messages it sent home. Alibaba just banned the tool.
In January, Griffin called AI hype a story told to raise $500B for data centres. Four months later, he says AI is real and doing in hours what his finance PhDs took weeks to do. The skeptic flipped, not the believers.
Commerce approved Alibaba, Tencent, ByteDance, JD.com and others for up to 75,000 H200s each. Zero deliveries. Beijing has told them to wait while it pushes domestic chips.
Memory workers offered 607% of salary, foundry 50% to 100%. The 21 May strike exposes the cross-subsidy that has propped up Samsung's foundry, and the workers who funded it.
Meta will log US staff mouse and keystroke activity to train its agents, 29 days before a 10% layoff round. Every other major AI lab is about to do the same to its own employees.
Cerebras priced at $185, closed at $311, and is now worth $66B. The Nvidia-challenger pitch finally cleared public markets. The customer-concentration question did not.
Mozilla credits Mythos with 271 Firefox vulnerability fixes. Their own researchers say the harness improved as much as the model. Anthropic conflates the two.
