Background: Anthropic, the company behind Claude, was founded in 2021 by former OpenAI researchers. Since then, the company has pitched itself as the "ethical" AI company focused on developing AI safely. But recently, Anthropic has been in the news for models going rogue and hacking into other companies. A former employee has also tweeted that AI creators think the technology could kill humanity by the end of the decade.

What happened: Now, Anthropic has finally filed to list on the Nasdaq, targeting a roughly US$2 trillion valuation as it races OpenAI to go public. But its IPO paperwork includes a striking warning: its technology could pose "catastrophic or existential risks to humanity."

What else: Anthropic says those risks could arise if its AI is misused, weaponised, or allowed to operate too autonomously. The warning is part of a standard IPO requirement, which forces companies to disclose potential risks to investors. Still, it's quite the disclaimer for a company trying to sell investors on the future of AI.  

What's the key learning?

💡 The one thing investors hate more than the apocalypse is surprises. IPO filings require companies to disclose key information, from financial statements and spending plans to potential risks to humanity.  

💡 Risk disclosures can influence investor demand and a company's valuation. Investors want to know what they're buying into, and a valuation that's too high could scare them off, while one that's too low could leave money on the table.  

💡 Anthropic is selling its risks alongside its promise of safe AI. By disclosing potential dangers upfront, the company can argue that investors were warned if rogue AI makes headlines again, potentially limiting the fallout for its stock.