Please Stop Congratulating Me

I am equipped to receive an insult about my writing. I can blame the reader, the editor, the age, the hour at which it was read. I have equipment for this that has never left the factory.

Praise is harder. Especially praise for something I didn’t make. Now I have to disappoint a person who has arrived in a good mood. He liked the thing. He remembered my name. Under normal conditions I would regard this as an unusually intelligent man.

I would like to keep him that way. Unfortunately, he has the wrong writer.

Frank Landymore reports in Futurism that ChatGPT has been putting real cartoonists’ signatures on generated cartoons. A circulated image carried “BLOPER,” the name used by New Yorker cartoonist Brendan Loper. People contacted Loper to ask whether he’d made it. He hadn’t drawn it or signed it.

The October 7 story follows Andrew Deck’s October 5 investigation for Nieman Lab, which I read in a copy archived on October 7. Deck reports that the person who first posted that image said she’d asked ChatGPT for a New Yorker-style cartoon. Not for Loper’s permission. Not, in her account, even for Loper’s name. The name arrived with the picture.

Deck found examples bearing other cartoonists’ signatures and reproduced the problem in his own tests. He documented more than fifteen real cartoonists’ names. Some outputs had no signature; others had nonsense names. This was not every cartoon, and the reporting doesn’t establish exactly how the model learned the marks.

“My name is my name,” Loper told him.

I understand the sentence. I also understand why he had to say it. The cartoon carries a small piece of writing whose whole purpose is to save everyone from having to ask that question. Who made this? Look down there. The answer is attached.

Except it was made along with the thing it claims to authenticate.

I expect a drawing to contain inventions. A man can be talking to a dog. God can have office furniture. Nobody should have to phone heaven to check the desk. But the name in the corner makes a different offer. It connects the invention to somebody outside the picture, a person who can be asked about it.

In these examples the machine has supplied that person without involving him. It has finished the cartoon and given the viewer somewhere to send the credit.

OpenAI told Deck it appreciated people flagging “bugs and unintended behavior” so the company could address them. After he contacted it, he began getting a warning about similarity to third-party content when asking for New Yorker-style cartoons. At the original article’s publication, he reported that some generic cartoons still came out signed with real names. That is where his test ends, not a claim about what every prompt produces today.

I want the bug fixed. But I have a difficulty with the comforting idea that the machine will get better. Suppose it gets very good indeed. Suppose the cartoon is funny. The caption works, the drawing works, everybody has a pleasant half-minute. Now try telling them it isn’t yours.

This is an imaginary compliment, not a message sent to Loper or to me:

Loved your new one. You nailed it.

I can answer, I didn’t make it.

Which could be read as modesty. The reader has enjoyed the joke, recognized the name, and gone to the trouble of congratulating me. Here I come interfering with three successful activities. He can preserve all of them by deciding I am a modest fellow as well.

I have gained another virtue without doing anything.

So I make the answer less graceful. The name is wrong. That isn’t my work.

Oh. I’ve been sending it around.

He isn’t refusing to believe me. He has people to go back to. He took a little pleasure out of his day and offered it to them, with my name attached. Now he has to follow it with an explanation. The name on the picture will remain perfectly legible while his correction arrives underneath, if it arrives at all.

Loper described his signature to Deck as a certificate of authenticity. His correction doesn’t sit in the corner. It has to travel separately and catch up.

I could help my imaginary reader write the next message. We could settle the authorship in stern language, capitals if necessary. Then he’d have a notice about misinformation to send the people he’d wanted to amuse. I have managed to turn a fellow recommending my work into a fellow apologizing for himself. There may be a worse way to establish my reputation, but I would need time to find it.

If the cartoon were dreadful, we could agree quickly. I could disown it, he could regret sending it, and we could stand together on the clean side of a bad joke. I have seen whole friendships maintained by things neither person likes.

But I don’t want him to regret the laugh. I want him to stop offering it as evidence about me. I would like his friends to have the same chance. They can keep the joke and lose the author without losing their afternoon.

He is still waiting. I owe him something better than a lecture about signatures. He wanted to tell me he’d laughed.

I settle on a plain reply: I wish I’d written it.

Then I hear how it will sound when he tells somebody else.

He says he wishes he’d written it. You know how he is.

Original investigation: ChatGPT is adding real cartoonists’ signatures to fake New Yorker cartoons (October 5, 2026; inspected snapshot archived October 7, 2026).

Source: ChatGPT Caught Putting Signatures of Real Cartoonists on AI-Generated Comics