Meta-owned WhatsApp has recently initiated testing for a novel feature involving AI-generated stickers. Within the Android WhatsApp beta program, select testers have identified the emergence of AI-powered stickers, enabling users to craft stickers based on textual descriptions.
AI Generated Stickers
Although the specific generative AI model employed by WhatsApp remains undisclosed, according to reports from WABetaInfo, the stickers are produced using a secure technology provided by Meta, the parent company. This innovation appears reminiscent of how models such as Midjoureny or OpenAI's DALL-E generate images from text input. The forthcoming WhatsApp functionality is anticipated to allow users to generate personalized images for sharing as stickers with their WhatsApp contacts and groups.
The introduction of AI-generated stickers raises valid concerns about the nature of images that could potentially be created and shared on WhatsApp. Given the platform's history of grappling with the proliferation of misleading content during significant events like elections, incidents of violence, and the COVID-19 pandemic, precautions are paramount. While the ability to report inappropriate or harmful stickers is present, the safeguards implemented on the AI model's side remain ambiguous.
Reports suggest that stickers produced through this novel AI-driven feature possess distinct characteristics, suggesting the possibility of watermarking or indicators denoting their generative AI origin. Comparable efforts by other platforms, such as Microsoft's integration of OpenAI's DALL-E for Bing Chat users, have included conspicuous labels to denote AI-generated content. Similarly, Instagram is purportedly developing labels to identify content crafted with AI models