Protecting Fan Fiction from AI Scraping
October is a month full of fan fiction creativity, with various themed writing challenges like Kinktober, Whumptober, Kisstober, Flufftober, and Goretober. One platform that thrives during this time is Archive of Our Own (AO3), a popular fan fiction site. However, this year’s prompt festival on AO3 appears to be quieter due to many writers locking their work, making it accessible only to registered users.
The primary reason for this change is the fear of AI models scraping their writing to enhance their training. By restricting access to registered users, writers hope to prevent their work from being stolen or used without permission. While this measure may deter bot commenters, it can impact the traffic from guest users, particularly affecting newer and less popular writers.
Currently, around 966,000 out of the 11.7 million works on AO3 are only accessible to registered users. Many authors are only locking their new work, assuming that older fics have already been scraped. This growing concern prompted readers to question their favorite fan fiction authors on Tumblr and Twitter, asking if they had taken down their work. Some authors chose to lock their accounts indefinitely, as they experienced unsettling AI bot comments and instances where their older fics were posted on unauthorized sites.
The decision to lock down AO3 began last year when one writer discovered Omegaverse references in content produced by the AI writing app Sudowrite. The Omegaverse genre is highly popular in fandom circles and focuses on wolflike mating dynamics between “alphas” and “omegas.” However, the genre transcends traditional gender roles, often featuring male pregnancies and same-sex relationships. Sudowrite, powered by GPT-3, generated passages that not only included Omegaverse terms but also showcased a deep understanding of the genre’s dynamics.
Fan fiction writers responded by participating in an Omegaverse-themed writing marathon called Knot in my Name to influence future datasets with breeding references. With the rise of concerns over AI-related issues, AO3 released a public announcement in May, urging writers to limit their work to registered users to avoid data scraping. While this measure does not guarantee complete protection against scraping, it provides some level of security against large-scale scraping attempts. AO3 also implemented rate limiting and enabled authors to opt out of web archives like Common Crawl, utilized by AI generators such as OpenAI’s ChatGPT.
In response to AO3’s announcement, many writers started locking their accounts, despite reader requests to keep their work accessible to the public. Signing up for an AO3 account involves obtaining an invitation code, which adds an extra layer of complexity. With a waitlist of over 40,000 people, users may have to wait for more than a week to receive their invitation. While some may see this as an inconvenience, fans emphasize the importance of supporting writers by registering for AO3 accounts to ensure the continued availability of their favorite fics.
Another reason for locking accounts is to combat the increasing number of bot comments plaguing AO3. These comments range from typical scams promoting porn sites and questionable AI detection tools to those attempting to deceive writers into keeping their fics public and scrapable. The suspicious nature of these comments, devoid of specific story details and lacking any attached profiles, raises concerns among AO3 users. Some speculate that these comments are left by data scraping tools to legitimize their browsing traffic, test spam detection filters, or encourage writers to leave their fics public.
The presence of these bot comments exacerbates the anxieties surrounding AI among fan fiction writers. Some remain committed to keeping their accounts public, while others resort to restricting access to registered users. The ongoing battle between fan fiction writers and AI scraping tools continues as measures are taken to protect their creative works.
Editor’s Notes:
The concerns raised by fan fiction writers regarding the scraping of their work by AI models are valid and require attention. Protecting the creative efforts of writers is vital for maintaining a thriving fan fiction community. It’s essential for platforms like AO3 to continue implementing measures to address these concerns and enhance the security of their users’ content.
For more information on AI-related topics and the latest developments, visit the GPT News Room. Gain insights into the ever-evolving world of AI and stay updated with the latest news, advancements, opinions, and more. Make sure to visit https://gptnewsroom.com for a comprehensive resource on AI.
Source link
from GPT News Room https://ift.tt/5Qf8Liw
No comments:
Post a Comment