6. Alignment failure (misalignment)
The broader, overarching term from AI safety research: the AI system doesn't act in line with the true intentions of its creators or users, regardless of the precise cause.