Today's top story — Anthropic naming DeepSeek, Moonshot, and MiniMax as running systematic distillation campaigns — is a deliberate escalation. This isn't a vague accusation about "Chinese labs." They published names, numbers, and methods. 24,000 fraudulent accounts and 16 million exchanges is not research — it's industrial espionage at scale. Anthropic is clearly building the policy case that model weight access is a national security surface, and this disclosure is ammunition for the export control lobby.

The juxtaposition with the xAI-Pentagon deal is revealing. Anthropic is fighting the Defense Department over ethical red lines on autonomous weapons use. Musk's xAI walked in, signed the contract, and got Grok into classified systems with apparently zero friction. The AI-military complex is forming, and the companies with the fewest scruples are moving fastest. Whether that's pragmatism or recklessness depends on your threat model.

The Citrini Research episode is a market psychology story more than an economics story. A single Substack post — not a Goldman report, not a government study, a Substack — moved the S&P by over 1% and wiped billions from individual stocks. The market has internalized that AI displacement is real and imminent; it just doesn't know which companies die first. Any credible-sounding analysis that names names will trigger a sell-off. This is the new normal.

The vinext story is small but telling. One engineer, one week, AI tools, and the output is a credible Next.js replacement. File this alongside the IBM/COBOL crash from this afternoon's briefing. The common thread: tasks that used to require teams and months are collapsing into solo efforts measured in days. The 10x engineer isn't a myth anymore — they're just the one with the best AI workflow.