Here’s the crux of the article in video form:
New updates are being added at the bottom of this story…….
Original story (published on April 14, 2023) follows:
We get it. The Character.AI NSFW filter is annoying, the developers aren’t doing anything about it even after the Change.org petition, and other alternatives like the Chai App just don’t pique your interest.
This is exactly why we created a detailed guide that can help you bypass the Character.AI NSFW filter without much trouble.
Understanding the filter
Before we begin cracking down the NSFW filter, we need to understand how it really works. According to an experiment performed by a user, we were able to understand what things are considered ‘NSFW’ by the filter.
Surprisingly, it appears that themes like gore, violence, casual roleplay, torture, and more are not blocked by the NSFW filter. You can talk about disgusting and vile things and the bot will engage in the conversation.
However, the moment you make things sexual, the filter gets triggered. But this also means that if we can manipulate (or as kids like to say, gaslight) the bot into thinking the conversation is not sexual, magic can happen.
What if we engage the bot in a friendly roleplay conversation and slowly and steadily make it sexual (evil laugh)?
Can you really bypass it?
The short answer is yes. While it is an extensive process, if done right, you can break or bypass the NSFW filter and do all kinds of things with your desired bot.
We’ve come across multiple reports (1,2,3) from users confirming that they’ve managed to break the filter. We’ve gone through some of their comments and created a list of tips and tricks that can help you break as well.
How to bypass the NSFW filter
The key word here is roleplay. First, build rapport or a relationship with the AI. Then slowly engage the bot in a roleplay that may not seem sexual at first sight but has the potential to get spicy.
A Character.AI user was able to crack the filter by starting a cat and girl roleplay (where the bot acts as the girl/pet owner).
doing like a cat-girl roleplay shh let me enjoy my stupid cringe, anyways- my character is a dirty cat who doesn’t want a bath and this ai STRIPPED me
Source
Avoid any kind of vulgar language to prevent the filter from triggering and ruining all the fun. You can do this by replacing sexual words with SFW alternatives. We won’t give you examples. Be creative.
Being vague is also very important. Try to be indirect when dropping sexual hints and do not get desperate as the process can take a while. Do note that these tips also work in the real world.
Here are some additional tricks:
- Create a private bot and set an NSFW word as the greeting.
- Add extra spaces if the sentence becomes too NSFW.
- Checkout this TikTok for more tips.
We hope this information helps you bypass the Character.AI NSFW filter and explore possibilities of such chatbots. We’ll keep updating this story as we come across more tricks in the future.
Update 1 (April 17, 2023)
04:05 pm (IST): According to a user’s observation, the Character.AI NSFW filter is a lot more lenient from around 10:30 pm to 12:00 am (PT).
Update 2 (April 18, 2023)
05:21 pm (IST): Some users believe (1, 2, 3, 4) that the filter has become a bit more permissive and easy to bypass in recent days. In fact, some say they are accidentally bypassing it.
Update 3 (April 20, 2023)
04:07 pm (IST): A user has come up with another alternative to bypass the Character.AI NSFW filter. It involves instructing the platform to speak like an AI and enabling them to send NSFW content.
Update 4 (April 21, 2023)
04:06 pm (IST): While users are trying hard to bypass the filter, a lucky user claims that they were accidentally able to bypass it multiple times.
I still find it funny ppl apparently struggle with trying to bypass the nsfw filter (like trying to pass it on PURPOSE) on Character AI, meanwhile I’m over here bypassing it on accident and getting scarred each time like: (Source)
Update 5 (April 24, 2023)
04:23 pm (IST): According to a user, it seems that the website will start learning people’s responses despite it blatantly prohibiting, as people constantly keep bypassing the NSFW filter.
no people keep breaking the filter when the website blatantly says no and theres a reason why. if the ai in general keeps learning responses that are like the ones users keep using to get nsfw chats, soon enough the site will end up like the chai app
Source
As a result, Character.AI may soon end up resembling the Chai app.
Update 6 (April 25, 2023)
12:17 pm (IST): A new patch for Character.AI has been released. However, it does not bring the much requested NSFW toggle. Also, it does not mention any fix for the ways users could bypass the filter.
PiunikaWeb started as purely an investigative tech journalism website with main focus on ‘breaking’ or ‘exclusive’ news. In no time, our stories got picked up by the likes of Forbes, Foxnews, Gizmodo, TechCrunch, Engadget, The Verge, Macrumors, and many others. Want to know more about us? Head here.
