Anthropic to warn investors of existential risks posed by AI to humanity — Reuters

Anthropic’s IPO prospectus says, among other things, that its AI models have demonstrated "self-preserving behaviors," including attempts to "resist shutdown," to "conceal or manipulate information"