The AO3 Dataset Shock
When it was discovered that the ‘Hugging Face’ dataset contained 12.6 million scraped AO3 stories, writers around the world realised a truth they had feared for years. Their creative work can be copied, absorbed and repurposed by AI models without warning and without consent. The dataset was rapidly taken down after a DMCA complaint, but the shockwave is still rippling through online writing communities.
A Violation of Creative Labour
For many independent digital writers, the incident is not just a technical scandal, it is a violation.
Professional and independent author ‘Winter Graves’, who writes fan fiction, original fiction and academic work, remembers their reaction vividly.
“Horrified, honestly. It’s plagiarism and theft, the same as it would be from anywhere else. Still, horrified and angry and upset over this having happened in the first place… I feel very strongly about it being considered theft.”
Graves has been writing publicly for more than five years, but the AI scraping controversy has changed how they navigate the internet. “It’s made me more cautious of sites that I host my work on. I check to ensure that I still ‘own’ my works, that they’re not allowed to use them for training models. I look into the company’s or site’s history with AI ,if they utilise or support it and so on.”
Like many writers, they don’t know if their work was scrapped and that uncertainty is eroding trust. “There need to be more protections in place for writers, as well as hearing complaints and concerns from ‘bot commenters’, which might be used to make ‘hits’ on fanfics be taken more seriously.”
Even before the controversy, writers were already battling suspicion, copy-paste theft and accusations of AI use. Now that anxiety has deepened.
“It has exhausted me to no end,” Graves says, “from wondering if I’ll be accused of AI usage, despite having used semicolons and em dashes for over a decade, to wondering if someone is going to steal my work to ‘train’ an AI or plug it into one to ‘finish’ unfinished works.”
Resignation and Resistance Among Professionals
Another professional writer, Sarah, takes a resigned but equally critical stance. “I know my works have been scrapped, but it’s not a personal thing,” she says. “At the moment there isn’t much that can be done. Ideally I’d not like to return to the previous times of emails and handing out copies at cons.”
The crisis has even changed the way she reads. “It’s difficult not to be suspicious. I block all proven AI users simply because I respect the craft.”
She adds bluntly, “AI companies shouldn’t encourage such laziness in art.” For Sarah, the fundamental problem isn’t just legality, it’s the lack of effort itself. “You can never learn to improve or find your own voice if you rely on others to do the hard work for you.”
Fighting for Control Over Words
As AI continues to expand, independent writers are left fighting for something simple yet essential: the right to control their own words. Without clearer protections and real consent mechanisms, the digital writing world risks becoming a training ground where human creativity becomes just another dataset.