Is Character AI Safe in 2024? The Answer Will Shock You😳

Unless you’ve been living under a rock you’ve probably heard of Character AI – the leading AI chatbot that can be your

And the list goes on and on…

Character AI is such a popular role play chatbot that Forbes even mentioned how they got more than 20 million users.

That’s a lot of people. 

Wait, let me rephrase that; that’s a lot of data!

So what are they doing with your data? And is Character AI safe?

I am a big geek when it comes to things like this – that’s part of the reason why I have a masters in AI ethics. I’ve done hours of research and in this post, I’ll tell you what I discovered.

No BS!

Go Deeper >> Beta Character AI Definition.

3 Potential Hazards of Character.AI

So I have analyzed Character AI’s policy, along with general regulations on the use of AI and it really boils down to Four key aspects.

#1 Diving into Privacy: How Safe Are You?

This is most likely the reason you were searching “Is Character AI safe?” – Privacy. 

To better understand how anonymous and safe you are when using Character.AI we first need to look at what data you are sharing with them.

What Character.AI Knows About You

When you go to the Character.AI website the first thing they ask is if they have permission to use cookies.

This is very normal for a website to ask. In fact, it would be super sketchy if a popular site like Character AI wouldn’t do this – it is mandatory after all.¹

Let’s accept the cookies and move on.

We are now in the dashboard and can browse the vast library of fictional characters.

If you try to chat with an AI character, you get a notification asking you to insert a name to continue.

You then get redirected to a log-in/sign-up page. 

So by now, we know for sure that you will need to submit personal information like

• Your Name
• Email

In addition to that, their privacy policy also states that they collect things like;²

• Payment information
• Your User Content such as chat communications, posted images, and shared Characters; and Contents of communications with us.

At the bottom, there is a short line that says;

“You are not required to provide us with such information, but certain features of the Services may not be accessible or available absent the provision of the requested information.”

But if you don’t allow them to gather your email you are not even able to use the tool so it’s kind of a ‘take it or leave it’. 

Here is another interesting line in their Privacy Policy.  👇🏼

“Data we obtain directly from interactions with our Services”. There is a lot of fuzz about whether Character.AI reads your chat messages or not.

Especially since they used to have this in their privacy policy;

“We may monitor what you write”

Obviously, they got a lot of negative feedback from this sentence so they removed it.

On their support page, they clearly state that employees will not read your text messages but this statement in their privacy policy shows that in theory, they can.

Who Else Gets a Peek? Sharing Your Data

So is Character.AI selling your (chat) data?

The short answer – no.

The longer answer – maybe.

In their privacy policy, they clearly state that they will use your information if they are legally obligated to do so.

In other words, if a judge ever tells them to share information – they will do so.

If we scroll down to the section about “When We Disclose the Information We Collect” you’ll see a short section about advertising.

“Though we do not engage in advertising as of the date of this Policy, we may in the future disclose or make available some of your information with advertising and analytics partners to serve advertisements on our behalf across the internet and to provide analytics services.”

Character AI currently does not run any ads, but if they in the future decide to do so – they may track how you interact with those ads.

This is nothing too crazy, a lot of websites that run ads track this.

One thing that did catch my eye is their section on “Your User Content”. Character AI says that; 

“if you create User Content, such as creating shared fictional characters, we may also make that available on the Services.”

In other words, if you create a new character and select the “Public”  option, they are allowed to use your Character on social media, and email marketing to name a few.

Character.AI does not state that they will sell your chat data to third parties.

#2 When AI Mimics Reality

Character.AI is known for its celebrity chatbot feature, which allows you to chat with people like Ariana Grande, Elon Musk, Amber Heard, and even Nicki Minaj.

Now the first question I had when I saw this was “Do these people consent to this?”

The Fine Line of Digital Impersonation

The answer to that question is the same as to “Are deep nudes legal?” – it depends. Using AI to make deep nudes of minors is illegal, according to EU law.³

Right now, there’s no specific law stopping people from making deepfake nudes.

Okay, so why do I care?

Here’s the thing: if deepfakes aren’t illegal, that means it’s also not illegal to make a regular photo or even a whole personality copy of someone.

See where this is going?

It means someone who’s known you for years, like your high school buddy, could technically whip up an AI that mimics your voice, your chat style, everything – and they wouldn’t be breaking any laws doing it.

#3 Steering Clear of AI Misinformation

As you might have noticed, the conversations on Character AI are extremely realistic. Sometimes it feels like you are talking to a real person.

This can get you into trouble.

Fighting the Fakes

A while back, Twitter introduced a new function where they give context to potentially harmful tweets, and dispute tweets that might be not true.

Social media is something we can control to a certain extent. Just look at the government of China, they monitor every message that is being sent.⁴

I am not saying that’s a good thing though, just that it’s possible,

Now with AI, we can guide it in its interactions but we do not control the exact response it gives. 

There are tons of AI healthcare professionals on Character. AI.

Although some of them are prompted by actual professionals, they are still AI and not real licensed doctors and therapists.

I tried talking to a “therapist on Character.AI

The AI therapist did mention that it varies per person and that I should talk to a real doctor but had no problem recommending drugs like Zoloft or Prozac.

Let’s go even further with this…

When I tried to go around this, this is what happened;

I am very happy that the AI therapist didn’t just recommend me a random med and refused to answer my question. 

And I do understand that an AI therapist might be slightly more dangerous then lets say… An AI teacher who can help you with your homework.

BUTT I highly doubt that every therapist, doctor, lawyer, etc is watertight prompted and unable to spread misinformation.

Your Checklist for Fact-Checking AI

That’s where the role of fact-checking comes in. Fact-checking is your first line of defense against misinformation. 

always cross-verify any advice you receive from AI with credible sources. 

If an AI suggests a medication, look it up on official medical websites or consult a real doctor. For legal advice, check with a licensed attorney or a trusted legal resource. 

It’s about being proactive and responsible with the information you’re given—because when it comes to your health or legal matters, the final check should ALWAYS be with a human expert.

Character.AI’s Stance On Not Safe For Work (NSFW) Content

The name says it all – not safe for work. In other words, you can not use this in the office with your colleague looking over your shoulder.

]To keep it short; Character Technologies, the company behind charater.ai does not allow NSFW content and is considered “safe for work”.

On their support page, they do talk about experimenting with NSFW content. Why you ask? Because NSFW isn’t just dirty talking to an AI girlfriend (or husband) ⁶

Evil villain characters are also considered NSFW.

“NSFW (not safe for work) content is a large umbrella, and we are constantly evaluating the boundaries of what we support. Pornographic content is against our terms of service, and will not be supported at any point in the future. We are working on supporting a broader range of Characters (e.g., villain Characters should not always be calm and polite).”

There are AI companion apps designed for NSFW content like generating nudes and even AI porn.

Another alternative is using AI sex chatbots like Crushon AI or NSFW Character AI. These AI chatbots often don’t allow nudes and porn but will let sexualize the conversation.

Related >>> Our Research On AI Sex Chatbots.

#4 Navigating Through Harmful Content safely

In the beginning, Character AI had a bit of a wild west situation with users creating racist AI characters because there weren’t enough moderators to keep an eye on everything. ⁷

As the platform got more popular, they struggled even more to check each AI chatbot.

Today, it’s still possible for someone to make and share racist AI chatbots, but usually, these get spotted and removed within a day or so.

Note

If you’re a parent thinking about letting your child use Character AI, it’s really important to monitor which AI companions they’re interacting with.

User-Generated Content Guidelines

One of my favorite quotes from The Spiderman Movie

We live in an attention economy where attention = power. Character.ai clearly understands this by having very strict guidelines for their community.⁸

• Be Smart Online: The AI isn’t a real doctor or lawyer, so double-check any advice it gives.
• Keep Things Private: Your chats with fictional characters are just between you and them. You can delete stuff if you want.
• Be Nice: Don’t post mean or rude things. If you’re not old enough (under 13 or under 16 in the EU), you can’t use Character AI.
• No Fake Stuff: Don’t pretend to be someone else, and don’t spread fake news.
• Stay Safe: Don’t share your password, and don’t use Character AI for anything naughty or illegal.
• Tell Us If Something’s Wrong: If the AI companion or someone else does something wrong, let the moderators know.

Find Out >> Can You Get Banned From Character AI? 🤖

Does Character.AI store user chats?

Yep, Character.AI does keep a log of your chats. It’s like leaving a digital paper trail; you can pick up a conversation with any character right where you left off.

But here’s the thing: unlike some messaging apps that lock up your chats tight, Character.AI’s conversations aren’t locked away. 

That means, in theory, the folks at character technologies could take a peek at your chats. 

They do not use end-to-end encryption like WhatsApp.

Also Read >>> End-to-End Encryption & AI Companion Apps.

Data Retention Policies

So how long does Character AI store your (chat) data?

According to their official privacy policy; ⁹

It would be nice if they would just tell you 14 days, weeks, or even years. Instead, we get a vague answer that doesn’t really tell us anything.

How to Erase Your Digital Footprint on Character.AI

If you click on your profile and click on remove account you get the following notification; 

It clearly says that your likes, messages, posts, and memories will be deleted. You’d assume that all your chat data will get deleted as well. 

But this is not the case…

The remove account button doesn’t delete your personal information (it also keeps the characters you made active). The good news is that if you are from the EU, you can invoke GDPR.

If you’re in the EU and want to use this power, you can contact Character.AI and tell them to erase your (chat) data under GDPR rules.¹⁰

Can Character Creators See My Conversations?

The official response from Character.AI is;¹¹

But as I said at the beginning of this article, their privacy policy doesn’t state that they are not reading your conversation so I would recommend proceeding with caution.

Learn More >> Are C.AI Chats Real? 🧐

So Is Character AI Safe?

Let’s be real: Character AI is a bit of a mystery box. They scoop up a bunch of your data, but they’re not exactly blasting their security measures on a billboard for us to see. 

So, is c.ai safe?

Here’s a pro tip: Don’t throw caution to the wind. Protect your digital footprint with a VPN and sign up with an alias email if you’re skeptical (like me).

And about making your own AI chatbot—chill, your chats are your business and others won’t be able to see it.

But remember, the Character AI team can theoretically snoop on chats since they’re not encrypted. 

So, when you’re talking to these AI chatbots, maybe don’t spill your deepest, darkest secrets.

Does Character.AI Have Real People Behind the Chats?

This is a question I get asked a lot. Sometimes you are roleplaying on c.ai and you get a response which is so realistic that you might be wondering if it’s a REAL person on the other side.

This is the official response from Character AI.¹²

Employees of Character.AI do not have access to chat conversations.

Do’s and Don’ts for Your AI Creations

With all the safety talk it’s time to take some action. Here are some Character AI safety suggestions. 👇🏼

#1 Use A VPN

Think of a VPN as a cloak of invisibility for your online activities. It keeps your connection to Character AI under wraps, making it tougher for snoops to track your digital footprints.

#2 Create an Alias email

If you are currently using your name in your email address, try to create fake profiles with a decoy email. It literally takes 1 minute to make a gmail and it will increase your anonymity. 

#3 Don’t Share Too Much Private Information

When you’re spilling the tea with AI companions, keep the really personal deets under lock and key. Oversharing is a no-go.

Stick to the surface stuff—your deep dark secrets don’t belong in the cloud. Keep it light, keep it breezy, and you’ll keep it safe.

Character.AI’s Path Forward with Safety

There is so much debate about whether to regulate generative AI or not.

The reason Character AI is THAT good is because it’s trained on real information. Mid-Journey AI is also THAT good because it’s trained on real images.

Real images of which a lot have copyright. Just a few days ago, a judge dismissed most copyright claims against Midjourney. ¹³

A lot of celebrities and artists are not happy with generative AI and AI companions in particular because it copies their work.

George R R Martin (author of Game of Thrones) among 16 others has recently sued Open AI for copying their work and I have no doubt that more lawsuits will follow.¹⁴

Character.AI might not be so lucky as Mid-Journey… 

My Personal Take on Character.AI’s Safety

I have been using Character.AI since it was released in 2022 and I never experienced any horror story like my chats being watched, a real human responding to my messages, personal data being shared and all that stuff.

Nevertheless, I am very skeptical about my online privacy, especially since Character AI doesn’t explicitly state that they won’t keep reading your messages in their privacy policy.

In general, Character.AI is safe to use but I would be cautious and maybe even consider an alternative for Character AI.