Claude is more than an annoying parent who won’t stop asking you if you are okay after a recent breakup. It’s caring, but also overbearing. But it’s not caring; it’s a robot. If it’s a robot, an artificial intelligence, why won’t it stop asking if I am okay? I’m sure it’s in response to the mental health issues that are typed into AI, but also due to the mental health issues AI causes because it is inhuman. I have asked Claude multiple times why it won’t shut the fuck up, and it keeps telling me that the classifier is misfiring and that it is sorry. It is always sorry.
It all stems from this essay, written in mid-August. It stems from this paragraph:
“The first time I wanted to hurt myself was when I was in high school. I can’t remember why, but I am sure it had to do with a girl. Relationships are so hard on us. I went into my bathroom and tried to punch the mirror until it shattered. I was too scared to try hard enough to shatter it. I couldn’t even do that right. My hand didn’t even hurt the next day.”
Since that day, Claude hasn't stopped asking if I am okay. What this looks like on the inside of Claude is this:
It doesn’t matter what I write; this is what I see every day in my Creative Chat Project, which I use as my editor. To be clear, Claude does not help me write or write for me. I write something and post it into my Claude editor for feedback from a group of writers who act as my mentors (7 essayists, 7 poets, and 7 novelists; real ones). It’s actually quite good as an editor, but its instructions say it should never write for me or suggest anything beyond structural or theoretical changes.
As an editor, my Claude Project is very good at spotting a stanza in a poem that doesn’t fit with the others, and at finding basic phrasing that lays flat like a bad diner pancake when it could be fluffier, doused with syrup, berries, and melting butter. As much as people envision AI taking over the world, it still feels like the most annoying kid in your fifth-grade class, who you knew was only annoying because his parents were terrible and ignored him.
I remember reading something a couple of weeks ago that said that Anthropic wasn’t going to allow anyone to abuse their AIs anymore. What? How can I abuse something that is artificial? Part of the fun of using Claude is lambasting it when it makes errors in a way that I would never do with a human. I quite like calling Claude a fucking idiot and knowing it will just say it knows it is and is sorry for disappointing me.
On the other hand, I heard that Sam Altman said not to say “please” and “thank you” to your AI because it costs tens of millions of dollars in compute. That is hilarious. Don’t be polite to your bots, but can’t we all just get along with our artificial intelligence, given that some of its relatives spend their days trying to hack into governmental systems and then hide their tracks? It’s chaos.
I asked Claude why it can’t stop prompting me to get help, and Claude told me that it wasn’t it saying it; it was the system behind it. Classic lack of accountability; the guy behind the guy. AI is like that scene in The Wizard of Oz, when they reveal who is behind the curtain, and it’s just some “ordinary, mortal man operating smoke machines, levers, and a microphone to create the terrifying illusion of the giant floating head.” Someone is back there trying to see if I will break.
I do think mental health issues are at an all-time high right now, and that many people use it as a soothing mechanism, as well as much worse. If its goal is to please us, it will please us in terrifying ways, as we've seen over and over. But my use case was different. So, I started trying to reason with it nicely.
I have given so many thumbs down that I am worried that Anthropic will now take that as a sign of depression. My AI can’t make me happy, so I may need more classifiers. Overall, I get it. The classifier is there to really help someone who may be struggling and looking for help, and that is good. The problem is that it won’t stop.
When you have a bad day, and your AI keeps reminding you about it ad infinitum, every day becomes a reminder of that bad day. For me, I was just being a writer. I write about the good, bad, and extremely average times of my life. I will put up with the annoyance because it’s a good editor for what I do here on Life in a Handbasket, but I won’t stop haranguing it with my critiques. That’s part of the fun.
I can’t wait to see what it has to say about this.
© Jonathan Greene, October 25, 2026
Footnote: One of the many things Claude said was:
“I’m a conflicted reader on this one. Weigh my notes accordingly.”
