Hi, I'd like to bring attention to an issue that may put Chyoa users at risk of being bilked out of money or worse. There is a user called "pixbk" posting chapters in their own story and others' stories which they claim to be created by msinc-llm, "a proprietary large language model (LLM) developed by Pixbk." They provide a link to their website where users can try the model very briefly and pay for more tokens if they like the output. Example: https://chyoa.com/chapter/Remote-Possibilities---Mothers-and-Sons.1708691 They claim this model generates stories with embedded images based on image prompts, though the version on their website (linked at the top of each chapter) uses text prompts. The website offers 1000 free "tokens" to "generate" a story or two to see what the model is capable of. If the user wants more tokens, they can become a subscriber. This also unlocks the ability to choose a scenario (rather than it being random) and the ability to vote on new prompt options. After having fiddled around with this website for a bit, I am 100% certain that it is fake. These stories are not being generated on the fly by an AI model. They are generated ahead of time and heavily edited by a human, or entirely human-written. AIs hallucinate, it's a fundamental flaw of the technology. The stories produced by this website are completely coherent every single time, with the closest thing to a hallucination being the odd mild grammatical error. Furthermore, the stories have images and animated GIFs embedded into them which are always completely relevant and cohesive with the story, always embedded in logical places, and always placed at frequencies that make narrative sense. This is NOT something that an AI is capable of, full stop. Finally, once I figured out how to bypass the arbitrary and extremely low "token" limit allotted for unregistered free users, I found that eventually, it starts repeating outputs. I got the same story word for word and image for image that I did twenty or so attempts before. That's hard evidence that it's not generating the stories, but rather pulling them from a set of premade stories. So what we have here is an unusual situation. Usually you hear about people passing off AI-produced work as being made by a human, but in this case we have someone passing off human work as AI. They're charging people money for "tokens" and for the right to vote on new "prompts," so at best they're scamming people by charging them for a service they aren't actually providing. Worst case scenario, they're harvesting credit card data for identity theft purposes. In any case, I strongly advise the administrators of this website to remove all links to this third-party site ASAP.
