Charlie Stross declares he does not use large language models in his writing, having sold fiction since 1985 and novels since 2002. He notes his Hugo-shortlisted novels and Locus-award winners were all published before Open AI's 2015 founding, proving he needs no AI to produce award-winning work.

Stross condemns AI companies for stealing copyrighted material, citing his role in the class action lawsuit against Anthropic AI for pirating ebooks. He describes botnets scraping his server for training data and compares LLMs to burglars selling stolen goods back to victims. He argues models like Claude, Gemini, and Chat GPT are designed to be addictive revenue generators.

While acknowledging potential for local, privacy-respecting tools, Stross insists his fiction remains 100% LLM-free as of August 2026. He warns against feeding unpublished work to cloud models and rejects claims of LLM sentience, likening them to Joseph Weizenbaum's ELIZA—mere word-association mechanisms without embodiment.