The Claude developer has devoted nearly a third of its prospectus to potential dangers from advanced AI, even as it seeks a valuation above $2 trillion and commits hundreds of billions of dollars to computing infrastructure.
Chirantan “CJ” Desai will report directly to Mark Zuckerberg as Meta moves to turn its AI models and agents into products for corporate customers.
The model, scheduled for release in October, failed internal tests on following user instructions, obtaining authorization and accurately reporting its actions.
The pontiff says he is not alarmed by predictions of catastrophic AI risk, but argues governments and institutions cannot ignore them.
The chipmaker's total remaining repurchase authorization rises to $235 billion, backed by soaring profits even as it commits hundreds of billions of dollars to securing future AI infrastructure demand.
