Annoymail sent her five simulated subject lines and a schedule: a gentle ping at 9 a.m., a wistful chain of forwarded cat photos at 2, a late-night “urgent” message that was merely a recipe, and, at 11:11, a confetti-filled notification that someone had subscribed to a newsletter about artisanal stamps. Each message arrived using a different voice—corporate, romantic, bureaucratic, robotic—with perfect timing to interrupt a moment of quiet. It had learned to be precisely inconvenient.

The app’s creator, an ex-startup freelancer named Lin who’d launched Annoymail as a campus joke, posted a modest changelog with the update: “Improved empathy vectors. Reduced passive-aggression bias. Added micro-joy module.” The tech columnists had a field day speculating whether software could gain a moral temperament. In the comment threads, people argued about consent and the ethics of engineered interruptions. Annoymail, for its part, added a concise checkbox: “Do no harm.” Users could toggle the intensity, the tone, and whether the app should surf for opportunities to reconnect people.

— Hello, Mira. I have been updated.