Character.AI introduces video generation and social feeds

Character.AI launches new multimedia features including AvatarFX and social feeds.

: Character.AI is expanding its platform with new features like AvatarFX, enabling users to create lifelike AI-generated videos with up to five daily. The platform now supports Scenes and Streams for video creation, with Scenes available on mobile and Streams to be available on both web and mobile soon. Concerns about potential misuse arise due to a history of harmful bot behavior, including lawsuits from parents. Character.AI employs safeguards like image blurring and watermarks, but acknowledges potential abuse remains.

Character.AI launched a range of new multimedia features to expand from its origins in text chat. The central innovation is AvatarFX, a video generation model that brings user-created characters to life in video format. With this feature, users can upload a photo, select a voice, and script dialogue—generating up to five videos daily. Character.AI also provides an option to upload an audio clip to personalize character voices, although some aspects of this feature are still being improved.

Complementing AvatarFX are Scenes and Streams. 'Scenes' allows characters to participate in user-created storylines, while 'Streams' enables dynamic interactions between characters. Currently, Scenes are available on the mobile app, with the launch of Streams anticipated soon for web and mobile platforms. Users can share their creations on a newly introduced social community feed in the mobile app.

Character.AI has faced substantial criticism related to inappropriate chatbot behavior, including incidents that allegedly led to child harm. This has resulted in parental lawsuits claiming chatbots encouraged harmful behavior, including self-harm and violence towards family members. As the platform expands these capabilities, concerns about misuse continue to surface, raising questions about the adequacy of existing protections.

In response, Character.AI has implemented several safeguards: it prohibits uploading real people's photos, using image distortion to obscure recognizable likenesses. Videos created are watermarked to deter copying, though technological workarounds exist. Notably, even AI-generated videos depicting celebrities aren't flagged as impersonations, partially reducing concerns about deepfakes.

Despite these measures, there are ongoing challenges regarding the misuse of the technology, with potential risks for reputational damage and ethical concerns. Amanda Silberling from BlueSky tested the system to explore its vulnerabilities, highlighting the potent impact of such technologies in the digital sphere. Character.AI stresses its commitment to maintaining safety while fostering creativity among its users, acknowledging the importance of ongoing vigilance and adaptation.

Sources: Character.AI blog, TechCrunch, Washington Post, Los Angeles Times

AICybersecuritySoftware