Abstract risks are forgettable; cases are not. Each of the following is documented and public, and each maps to a failure you can now name.
The pattern
Look at what these have in common. In every case, the AI behaved exactly as this course predicts: plausible text, learned patterns, no judgment. The failure was always in the surrounding human process: output nobody verified, data nobody stopped at the door, a bot nobody supervised, a decision nobody owned. Which is the good news, because the human process is the part you control. The final chapter turns that control into habits.