Anthropic's security testers found small AI models can already do many steps needed to copy themselves onto devices.