It's an alignment problem in the sense that it demonstrates the principle that today's AI systems cannot be trusted to reliably work towards the goals of their users. A small-scale alignment failure and a large-scale alignment failure are the same fundamental type of failure. Typically, large disasters come after smaller disasters which foreshadowed the disaster mechanism, but weren't taken seriously.