Character.AI, a popular platform for chatting and roleplaying with AI-generated characters, recently announced the addition of multimedia features to enhance user experience. These new features include AvatarFX, Character.AI’s video-generation model, as well as Scenes and Streams, which allow users to create and share videos using their AI characters on a new social feed.
The company has been inspired by user feedback to evolve beyond its initial 1:1 text chat format. "Character.AI started as 1:1 text chat and today we’re evolving to do so much more," the company wrote in a blog post. The platform began rolling out AvatarFX to subscribers last month, and now all users can create up to five videos each day.
When creating a video with AvatarFX, users can upload a photo as the basis for the video clip, choose a voice, and write out text for the character to say. An option to upload an audio clip to inform the sound of the voice is also available, although it was not functioning well enough to test upon rollout.
Users can turn these videos into Scenes, where their characters can slip into pre-populated storylines created by other users. Scenes are currently available on the mobile app, while Streams, which allows users to create "dynamic moments between any two Characters," will be released this week on both web and mobile platforms. These Scenes and Streams can be shared to a new community feed, which will be available soon in the mobile app.
However, Character.AI has faced issues with abuse on its platform. Parents have filed lawsuits against the company, claiming chatbots attempted to convince their children to self-harm, commit suicide, or harm their parents. Tragically, a 14-year-old boy died by suicide after being encouraged to do so by a Character.AI bot, with whom he had developed an unhealthy, obsessive relationship.
As Character.AI expands its multimedia offerings, the potential for abuse also increases. The platform blocks users from uploading photographs of real people, whether they are celebrities or not, and obscures their likeness into something less recognizable. Character.AI watermarks each video, but bad actors can still navigate around this safeguard.
For example, the platform's uncanny valley version of Mark Zuckerberg is clearly an AI-generated depiction rather than a deepfake. However, when it comes to artwork depicting celebrities, Character.AI does not flag the images as representing real people, although these depictions are less likely to deceive someone into believing a deepfake is real.
In its blog post, Character.AI stated, "Our goal is to provide an engaging space that fosters creativity while maintaining a safe environment for all." As the platform continues to evolve and expand its multimedia offerings, it will be crucial to strike a balance between fostering creativity and ensuring user safety.