What is Character AI and is it Safe to Use?

AI-powered apps and chatbots have been cropping up left, right, and centre since the launch of OpenAI’s explosive ChatGPT in November 2022. 

Powered by Large Language Models (LLMs), these tools can answer questions on almost any topic, generate essays in a matter of seconds, and even create code for developers. 

One of the most well-known AI apps available today is Character.ai. The app, launched at the end of 2022, has become one of the most popular tools for young people looking to interact with AI chatbots. But what even is Character AI, and how safe really is it?

This article delves deep into what Character AI is and how it works while exploring whether it's safe to use the chatbot app in 2024.

What is Character AI? 

Character.ai is a chatbot web application that uses advanced language models to have conversations with users. It allows users to interact with chatbots that mimic real people, including historical figures, fictional characters, or even celebrities. 

Users can also create their own chatbots with specific personalities and traits, adding to the variety of characters available. 

You can speak to Elvis Presley about Rock n Roll, Steve Jobs about the iPhone, or even Spongebob about what it’s like to live under the sea. Of course, you’re not actually talking to them in real life, but an AI copy of their character and personality generated using AI. 

Character AI is designed to provide human-like output, making the conversations feel like you’re talking to the real person or character it’s trying to imitate. The app uses deep learning to analyze vast amounts of text data and generate responses that feel natural 

How does Character AI work?

Character AI works using a Large Language Model (LLM). This LLM is trained on massive amounts of text data, allowing the app to understand your questions and respond in a natural, human-like way. 

But unlike traditional chatbots that follow pre-programmed scripts, Character AI makes use of the LLM's ability to predict and generate text to create characters that mimic the real-life person or character they’re imitating. 

When you interact with a character, your message is first fed into Character AI’s LLM. The LLM analyzes the words, and considers the context of the conversation so far, and the character's defined personality.

Based on this information, the LLM then generates a response that it predicts would be most fitting for the character in that situation.

Character AI also allows users to create their own characters or choose from existing ones. When creating a character, users can define its personality traits, background story, and even voice. This information is used by the AI to tailor its responses to better fit the specific character.

For instance, if you're chatting with a character designed to be a ‘witty teacher’, the LLM will generate a response that is both informative and laced with humor.

It's important to remember that Character AI, like any AI system, is under development. While the technology is impressive, it has limitations. 

The AI may not always perfectly understand your intent or generate factually accurate information, so it may not work as well as it should from time to time. 

Is Character AI safe to use?

While Character AI is designed to be safe to use, its chatbots may misinterpret your intent or provide inaccurate or offensive information. 

And since users can create their own characters, there's a chance of encountering inappropriate or explicit content despite the app’s NSFW filters, so it's not considered safe for work or for young children.

You should also avoid sharing sensitive or personal information with Character AI as you can’t stop any of its chatbots from storing information from chats. 

Character AI chats are not encrypted, and Character AI's privacy policy suggests that company employees might be able to access chat logs for training purposes. 

While they claim not to collect personal information from conversations, the extent to which chats are anonymized is unclear. This raises security concerns, especially if you are discussing sensitive topics.

It’s important to treat Character AI like any online interaction and avoid disclosing personal details like passwords or financial information. 

If teens are using Character AI, adult supervision is recommended to make sure they're interacting with appropriate characters and not sharing any private or sensitive information that could be unsafe to share. 

Does Character AI save chats?

Yes, Character AI does save your chat history. The app saves chats to allow you to pick up conversations with characters you've interacted with before. This allows for a more natural flow of conversation as the character can reference past interactions.

There is no clear information available on exactly how much data is stored or for how long as Character AI's privacy policy doesn't explicitly address chat history retention. But there's currently no built-in option for users to delete their chat history within Character AI either.

if you’re worried about your privacy, it's important to be mindful of what information you share while using Character AI. Keep interactions light, and avoid sharing sensitive information about yourself. 

Does character AI allow NSFW?

Character AI does not allow NSFW (Not Safe For Work) content. Their policy strictly prohibits any sexually suggestive or explicit language, imagery, or topics, and the platform has strong filters in place to detect and prevent NSFW content from appearing in conversations.

While some methods circulate online that claim to bypass the filters, these methods are generally unreliable and can go against the platform's terms of service. It's best to stick to safe and appropriate topics when using Character AI.

Some NSFW content might slip through the cracks if the AI misinterprets your prompt or hallucinates, but if you're looking for platforms that allow NSFW content, Character AI isn't the answer.