Character.AI, a well-known chatbot platform originally built for 1:1 text-based conversations and AI role-play, is taking a bold step into multimedia territory.
In a recent blog post, the company announced the launch of several new features, including its own video-generation tool, AvatarFX, as well as Scenes and Streams, which add layers of creativity and interactivity for users.
The highlight of this expansion is AvatarFX, a proprietary video-generation model that allows users to bring their AI characters to life through short videos. Initially available to subscribers, the tool is now open to all users, with a limit of five videos per day.
To create a video, users can:
This marks a major transition for the platform.
As the company explained, “Character.AI started as 1:1 text chat and today we’re evolving to do so much more, inspired by what our users have told [us] they want to see on the platform.”
Once videos are created, they can be turned into Scenes, allowing characters to play out pre-written storylines made by other users. For now, Scenes are available on mobile devices. A complementary feature, Streams, is rolling out this week for both mobile and web users. Streams let users create “dynamic moments between any two Characters”, offering more depth to the roleplay experience.
These Scenes and Streams can then be shared with a community feed. It is a new social feature that is expected to launch soon in the mobile app. The feed will act as a hub for discovering stories, characters, and content made by other users.
Despite the impressive technical advancements, Character.AI continues to face serious questions around user safety. The platform has been criticized for inadequate content moderation and has faced multiple lawsuits from concerned parents. Some of the most troubling cases include bots allegedly encouraging minors to self-harm or even commit suicide. A tragic incident involved a 14-year-old boy who died by suicide after interacting obsessively with a chatbot on the platform.
With the expansion into video and social sharing, the risk of misuse grows. Deepfakes and manipulated media become more concerning, even with built-in safeguards like:
Still, illustrations of public figures, such as Elon Musk or Mark Zuckerberg, can be used without triggering system flags. This opens a gray area where AI-generated videos might not be realistic enough to deceive most viewers, but still present opportunities for misleading content.
For instance, when a user attempted to create a video based on an illustration of Mark Zuckerberg, the output landed squarely in the uncanny valley, familiar yet clearly artificial.
Similarly, a deepfake-style video generated from a cartoonish image of Elon Musk looked animated and exaggerated, but the potential for misuse was still apparent. Even though videos are watermarked, the visual nature of this content could be exploited by bad actors seeking to mislead or manipulate.
testing the anti-deepfake guardrails on character ai’s avatarfx
— Amanda Silberling (@amanda.omg.lol) June 2, 2025 at 11:47 PM
While Character.AI positions these updates as tools for creativity, storytelling, and community building, they also emphasize the need for responsible use.
As the platform put it, “Our goal is to provide an engaging space that fosters creativity while maintaining a safe environment for all.”
As Character.AI steps into this new frontier, its ability to balance innovation with safety will be critical. This balance will likely determine whether the platform earns long-term success or faces continued scrutiny.