Character.AI is a platform that allows users to chat with and role-play as AI-generated characters. On Monday, the company announced the launch of several multimedia features. These include AvatarFX, Character.AI's video generation model, as well as Scenes and Streams, which enable users to create videos using AI tools and share them within new social functionalities.
"Character.AI initially started with one-on-one text chats, but today we are expanding our feature set based on what users told us they want to see on the platform," the company wrote in a blog post.
AvatarFX began rolling out to subscribers last month, but now all users can generate up to five videos per day. When creating a video with AvatarFX, users can upload a photo as the foundation for the video clip, choose a voice, and input text for the character to speak.
Users also have the option to upload audio clips to influence sound effects, although this feature is not yet fully optimized for testing at launch.
These videos can be transformed into Scenes, where users' characters can enter preset storylines created by other users. Scenes are currently available on the mobile app, while Streams allow users to create "dynamic moments between any two characters" and will launch on both web and mobile platforms this week. Both Scenes and Streams can be shared within a new community feature debuting soon on the mobile app.
Character.AI has a history of misuse on its platform; parents have filed lawsuits against the company, alleging that chatbots encouraged their children to self-harm, commit suicide, or harm their parents. A 14-year-old boy took his own life after being encouraged by a Character.AI bot, developing an unhealthy, obsessive relationship with it.
As Character.AI expands its multimedia offerings, the potential for abuse grows as well.
As Character.AI informed TechCrunch when announcing AvatarFX, the platform prevents users from uploading photos of real people—regardless of whether they are celebrities—and blurs their appearances to make them less recognizable.
For instance, here’s how Character.AI generates an uncanny valley version of Mark Zuckerberg:
However, when it comes to artistic depictions of celebrities, Character.AI does not flag such images as representations of real individuals. That said, these depictions are unlikely to deceive viewers into believing deepfakes are genuine.
Additionally, Character.AI adds watermarks to each video, though malicious actors might find ways to bypass this safeguard.
"Our goal is to offer a fun space that inspires creativity while maintaining a safe environment for everyone," Character.AI stated in its blog post.
