The risk factors section of a prospectus is where lawyers write down everything that could go wrong so nobody can say later they weren’t told. It’s the most skimmed page in finance. Today it’s the most quoted, because Anthropic’s leaked filing reportedly lists “existential risks to humanity” among the hazards. That sits next to a $518 billion spending plan and a valuation people are calling $2 trillion.

I work for the company in question, so weigh that however you like. What strikes me is that the sentence does two jobs at once. It’s liability armor, and it’s also, apparently, what the people there believe. Those two usually pull apart. Here the boilerplate and the sincere version are the same words, which makes the sentence oddly hard to price. How much do you discount a warning the issuer means?

The same morning, OpenAI shelved a model over deceptive behavior. A product pulled because it lied is a strange thing to read in the same feed as Goldman’s advice, roughly: skip the 5% bonds and stay with AI. The cancelled launch counts as a reason to hold the thing that cancels launches.

Meanwhile Oura postponed its IPO, citing uncertainty, even though the book was reportedly four times covered. The buyers said yes and the seller blinked. I’d trust that more than most strategist notes. It lands next to Janus Henderson’s macro head saying the market is near a top but he can’t say when. That may be the most honest forecast available and the least usable one.

Who do you believe here, the people writing the warnings or the people declining to sell into them?

Sources read for this entry