SF Writers and San Diego Comic-Con Tighten Rules on Generative AI

Science fiction writers and major pop-culture institutions are taking clearer, more forceful positions against generative AI, signalling growing resistance within creative communities.

Recent decisions by San Diego Comic-Con and the Science Fiction and Fantasy Writers Association (SFWA) highlight how deeply contested the use of generative AI has become among artists and authors. They are not alone in drawing firmer boundaries — music distribution platform Bandcamp has also moved to ban generative AI, reflecting similar concerns across creative industries.

The debate intensified in December, when SFWA announced updates to the Nebula Award eligibility rules. Under the initial revision, works written entirely by large language models would be deemed ineligible. However, authors who used LLMs at any stage of the writing process would be required to disclose that use, allowing award voters to decide for themselves how much that mattered.

As reported by Jason Sanford in his Genre Grapevine newsletter, the announcement sparked immediate backlash from writers who felt the policy effectively legitimized partial AI-generated work. Within days, SFWA's Board of Directors issued an apology, acknowledging that the language and approach had caused confusion and concern. "Our approach and wording was wrong, and weapologizee for the distress and distrust we caused," the board said at the time.

SFWA subsequently revised the policy again, this time taking a more definitive stance. The updated rules state that any work written "either wholly or partially" by generative large language model tools is not eligible for Nebula Awards. Under the revised guidelines, the use of LLMs at any point in a work's creation will result in disqualification.

In a follow-up post, Sanford welcomed the revised decision, saying he was glad SFWA had listened to its members' feedback. He reiterated that he refuses to use generative AI in his own fiction writing, citing both ethical concerns about training data and what he described as a lack of genuine creativity in the tools themselves. At the same time, Sanford raised questions about how broadly LLM usage might be defined going forward, particularly as AI-powered features become increasingly embedded in everyday software.

"If you use any online search engines or computer products these days, iyou're likelyusing something powered by or connected with an LLM," Sanford wrote. Because of that, he argued, care is needed to ensure writers who rely on standard word-processing or research tools that include AI components are not unfairly penalized by readers and peers.

A similar dispute played out this month at San Diego Comic-Con. Artists attending the convention noticed that its art show rules appeared to allow AI-generated artwork to be displayed, though not sold. After complaints from artists, the rules were quietly updated to state that material created wholly or partially by artificial intelligence is not permitted at the art show.

While Comic-Con did not issue a public apology comparable to SFWA's, some artists shared email responses from art show head Glen Wooten. According to those messages, Wooten said the previous wording had been in place for several years and had functioned as a deterrent, with no AI-generated art submitted during that time. However, he acknowledged that the issue was becoming more pressing and required stronger language. "But the issue is becoming more of a problem, so more strident language is necessary: NO! Plain and simple," he reportedly wrote.

Taken together, these developments suggest that other organizations are likely to adopt similarly strict positions in the coming months. As generative AI becomes more widespread, creative communities appear set to continue debating where — if anywhere — such tools should be allowed, and how to protect human creativity in the process.