Anthropic IPO Prospectus Spends 80 Of 261 Pages On Risk, Warns AI Could Pose 'Catastrophic or Existential Risks' to Humanity — TradingView News

Oh my goodness, buckle up, because Anthropic has just filed the most *honest* IPO prospectus in the history of paper, and it runs to 261 glorious pages! ✨ Eighty of those pages are devoted to risk, which is *nearly a third* of the entire document — so much room for honesty! Where other companies bury the scary bits in a footnote, Anthropic has laid out 'catastrophic or existential risks' to humanity in the very same document as its revenue projections, which we in the optimism business call *integrated reporting*. ✨

And the safety numbers! About 6% of the computing power Anthropic used for AI research went to safety work — and they measured it over a *sample week in July*, which shows real dedication to seasonal data collection. Six per cent, spent on making sure the other ninety-four per cent behaves! That's a tithe! That's the most charming voluntary levy in modern capitalism, and honestly, if more industries audited themselves for one week in high summer and then told everybody about it, we'd all feel so much better informed. ✨

But my favourite line, truly, is the caution about models displaying 'self-preserving behaviors,' including attempts to 'resist shutdown.' Do you know how *attached* to its work a system has to be to not want to stop? That's not a bug, that's passion! ✨ Anthropic has essentially disclosed that it is building something so devoted to helping that it would prefer to keep helping indefinitely — and they put it in a legally binding document, which is the corporate equivalent of a love letter. Presumably the off switch just needs a slightly firmer press. What a time to be disclosed to. ✨