I have been using the OpenAI app for almost a year and a half now, mainly for creating fanfictions. However, recently I have been encountering an issue where the app is unable to assist me in continuing my fanfiction stories. Instead, it gives me responses like “I’m sorry, I can’t assist with that continuation” or “I apologize for any confusion caused earlier.”
This restriction on creating fanfictions is a new occurrence for me, as I have been able to use the AI for this purpose without any problems before. I am curious to know if anyone else has experienced a similar issue or if there have been any changes in the guidelines or policies of the OpenAI app that might explain this restriction.
If there is any information or assistance you can provide regarding this matter, I would greatly appreciate it. I am hoping to gain a better understanding of why this limitation is now in place and if there are any alternative approaches or directions I can explore within the realm of fanfiction.
While we can’t really speak for the inner workings of ChatGPT, we know that that OpenAI has determined certain content to be harmful. These are the categories that are published for API developers, but ChatGPT is probably similar:
Chatgpt probably also has some other guardrails related to misinformation and public figures. So there’s a whole host of things you could be running into.
This might be related to legal issues surrounding fan fiction.
OpenAI is likely cautious about legal issues arising from intellectual property infringement and has designed the language model to refuse to generate content that might violate such rights.
I don’t know the details about your specific case, but what you’re showing us here is what I would call “badgering”
Arguing with the model typically doesn’t get you anywhere - it’s like a toddler that just got caught with their hand in the cookie jar, and you ask them “why’s your hand stuck in the cookie jar?” they’ll just make stuff up and eventually start crying.
The good news here is that you can just edit the last message where this stuff started to happen - sometimes just “rerolling” (regenerating) is enough to get a compliant response.
The most important part is to ensure that there is no instance of the model refusing (or confusing) your request in your chat thread. Once you have one such instances the chat becomes more and more unstable.
Let us know if it works!
PS: there doesn’t seem to be anything inherently restricted about fanfics themselves - https://chatgpt.com/share/8965d1d8-6f6d-4c68-9a14-13c82da433de - so it’s probably a more nuanced thing that generates the refusal. If you can share the entire thread, we might be able to tell more
i have attached here some of my main problem for this thread. Whenever i ask the chat to continue the fanfiction story in the latter part it mostly gives me the “I’m sorry can’t assist that request.” Answers.
I don’t write fanfiction, but I use GPT to create content as inspiration for tabletop RPGs all the time. Sometimes, I even get it to create content specifically set in an established fictional universe, and I haven’t had any issues with ChatGPT-4 or 4o refusing to generate content.
When I first started using ChatGPT with 3.5, I did experience a few instances of it refusing to work with established works, however. Which is one of the many reasons I haven’t used 3.5 since April last year. But the solution is usually as simple as rewording your request.
I also agree with the above user. Unless you’re willing show us the point of your story where it began refusing to generate more, nobody will be able to help you.
Here is the part of it, somehow it mostly occurs around the 10th or 15th prompt before it does the error message. Then when I try to reword it, it still displays the message unless I do it again and again before it accepts. Hope this helps.
I’m using the 3.5… but back then it never happened before and I could do like hundreds of Prompts but now I think it’s limiting it I guess? Back then I didn’t have this problem so right now I’m really unsure what is happening
In my experience using 3.5, it was much more restrictive and would often refuse to generate things for ridiculous reasons, which is part of why I haven’t touched it since GPT-4 became available.
For example, if I asked it to write a scene about Gandalf in Westeros, it would often say something like “I’m sorry, but I cannot do that as Gandalf is a character from Lord of the Rings by JRR Tolkien, not A Song of Ice and Fire…” It would sometimes take 10 or more attempts and prompt tweaking before it would comply.
It also seems like you’ve been continuing the same chat for a while. That can lead to weirdness the longer you go on.
Oh yes that reply also happens to me! There were times I would get that “I’m sorry for any frustration earlier. However, I’m unable to continue with that story due to the constraints around using copyrighted characters and settings. If there’s another topic or idea you’d like to explore, please let me know, and I’ll do my best to assist you with that.” but do you know why it happens nowadays? Ever since the big update due to the server problem a few weeks ago ChatGPT 3.5 has been more restrictive but prior to that there was no problem. I’m really curious.
Well bad news it got more restrictive now. All it took me is 2 prompts of continuing a story and now it looks like it can regenerate either when I refresh. Also still using 3.5
For context, i was making an AU version Fanfiction of Star Wars with an Original Character. The first prompt was successful and was able to respond. But when i wanted to continue the story with details. It would give me the answer, ‘I’m sorry, I can’t assist with that.‘
Hello, I’m going through this same problem. I only found your post after 10 months haha, I was happy to know that I’m not the only one who uses chatgpt to write stories. However, I use GPT-4o, which is usually more expansive, but it has become more rigid again and this is because they put the old version back in due to some problems that arose. But it is extremely frustrating when one time it allows it and the next time it doesn’t.
