The Hugo award-winning Science Fiction-focused Clarkesworld Magazine can receive over 12,000 submissions in just one year. Of course, that was before the proliferation of free online AI models that can write a dull, monotonous, though technically legible piece of fiction.
On Monday, Clarkesworld Magazine editor Neil Clarke tweeted that the company had closed all submissions, writing “It shouldn’t be hard to guess why.” Looking back over the past few weeks, it’s clear that fake spam submissions made using AI-based large language models has inundated the magazine’s editors with nearly 35 times the number of fake submissions as the same time last year. Clarke wrote that his magazine received 50 of these AI-generated submissions before noon on Presidents’ Day.
On Tuesday, Clarke said they do plan to eventually open up submissions again, but he elaborated that “We don’t have a solution for the problem. We have some ideas for minimizing it, but the problem isn’t going away.”
The problem he’s referring to is the incredible number of submissions the magazine has received in the last few months. OpenAI’s massively popular AI chatbot ChatGPT was released in November last year, and the number of submissions the magazine received into this month has increased exponentially since then.
In a blog post last week, Clarke wrote his company has never experienced plagiarism or artificially generated content anywhere on this scale. The Hugo-winning magazine had previously dealt with people making use of programs to insert different words or names into previously published work. In the first two weeks of February, Clarkesworld banned 38% of those new submissions, calling them spam.
“It’s clear that business as usual won’t be sustainable and I worry that this path will lead to an increased number of barriers for new and international authors. Short fiction needs these people,” the Clarksworld editor wrote.
Clarkesworld is one of the few major fiction publications that allows free fiction submissions, and Clarke said that restricting access to only those willing to pay to submit “sacrifices too many legit authors.” The editor bemoaned the lack of accurate AI detectors or any cheap tool to conduct any identity confirmation. These programs may be able to detect some AI-generated content, but Clarke is correct when he said these programs are “prone to false negatives and positives.”
He blamed the flood of fake content on “side hustle” promoters convincing people they can make a quick buck through AI-generated submissions, though it’s unclear which online hustlers are promoting that line of thought. Clarkesworld offers 12 cents per word for fiction between 1,000 and 22,000 words, so it’s not like anyone would make much, even if their submission was somehow selected for publication. Gizmodo reached out to Clarke for comment and for more about the problems he sees with AI-generated content, but we did not immediately hear back.
With the proliferation of generative AI, multiple other artistic industries are dealing with a deluge of artificially created content. Voice actors are trying to fight back against AI-generated voices stealing directly from their work. A trio of artists recently sued several AI art generator companies, alleging the training data used to create the diffusion-model AI scraped up thousands of copyrighted work, all without the original owners’ permission.