August 1, 2026
Em-Dash of Destruction
Charlie Stross – On the non-use of AI in my writing process
Award-winning author says AI can keep its hands off his books—and commenters brought knives
TLDR: Charlie Stross says he does not use AI in his writing and accuses AI firms of building products on stolen books while harming working authors. Commenters split between cheering his anti-AI fury, nitpicking his tech details, and turning the whole thing into a surprisingly vicious joke war about em-dashes.
Science fiction author Charlie Stross didn’t just say he doesn’t use AI to write—he torched the whole idea. In his post, he called today’s text generators glorified word-mashers, blasted companies for training them on stolen books, and reminded everyone that he was publishing acclaimed novels long before modern AI companies even existed. He also tied his anger to something very personal: his work is part of a legal fight over book piracy, and he says bot scrapers hammering his old server have made his site crawl. In short, this was not a soft “not for me” statement. It was a full flamethrower moment.
And the comments? Absolute feast. Some readers cheered the old-school author fury and basically said, yes, this man was warning us about weird future tech back in 2005 and now looks painfully right. One admirer called Accelerando “probably the best” fiction about the future explosion of intelligence, which made the whole thing extra delicious: the guy who wrote wild future-tech stories is now being called a “neo-Luddite” for wanting human-written books.
But not everyone was simply applauding. One commenter jumped in with a classic nerd correction, saying they hoped Stross wasn’t mixing up image-making AI with language AI. Another latched onto the post’s em-dash rage, joking that em-dashes have sounded pretentious since the 1890s and that AI has accidentally turned “serious writing” into self-parody. The result: part moral outrage, part pedantic fact-checking, part punctuation warfare—and very, very online.
Key Points
- •Charlie Stross says he does not use large language models in either his writing work or his personal life.
- •He cites a publishing record that predates OpenAI, including novels and award-winning works released before 2015.
- •Stross says he is part of the settlement in a class action lawsuit against Anthropic over pirated ebooks allegedly used to train LLMs.
- •He says his blog has experienced slowdowns and outages because botnets have been scraping it for data used in LLM training.
- •He argues that authors should avoid submitting story material to major LLMs such as Claude, Gemini, Copilot, and ChatGPT because prompts may become training data.