Snapchat on Tuesday announced new artificial intelligence (AI) tools for users. The social media giant revealed during its sixth annual Snap Partner Summit that it plans to introduce an AI video tool for those with creator accounts. The AI tool will allow users to create videos from text and image prompts. All videos created using AI are said to be watermarked by the company to ensure that other users can distinguish genuine videos from AI-generated videos.
In a press release, the social media company explained the new features in detail. The AI video tool is one of the most exciting features announced during the event. Dubbed Snap AI Video, it will only be available to creators on the platform. Specifically, to become a creator, users must have a public profile, be active posters for their stories and spotlights as well as have a large audience.
The feature appears to be similar to a typical AI video generator and can generate videos from text prompts. Snapchat said that soon creators will also be able to create videos from image prompts. This feature has been launched in beta on the web for a select group of creators.
A spokesperson for the company told TechCrunch that the AI feature is powered by Snap’s in-house foundational video model. Once the feature becomes widely available, the company also plans to use icons and context cards to tell users when a Snap was created using AI. A specific watermark will remain visible even when the content is downloaded or shared.
The spokesperson also told the publication that the video models have been thoroughly tested and security assessed to ensure that they do not generate any harmful content.
Additionally, Snapchat has also released a new AI Lens that lets users look like their elders. Snapchat Memories, available to Snapchat+ subscribers, will now support AI Captions and Lenses. Additionally, the company’s native chatbot My AI is also being improved and can perform many new tasks.
Snapchat says users can now solve more complex problems with My AI, interpreting parking signs, translating menus in foreign languages, identifying unique plants, and more Are. Finally, the company is also partnering with OpenAI to give developers access to multimodal large language models (LLMs) so they can create more lenses that can recognize objects and provide more context.
