Meta, formerly known as Facebook, has announced a plan to incorporate invisible watermarks into all images generated using its artificial intelligence (AI) systems as part of an effort to counter potential misuse of the technology. In an update about Meta AI, the company revealed its intention to implement invisible watermarks to all AI-generated images produced through the "Experience Imagination with Meta AI" tool. This watermarking aims to enhance transparency and traceability, making it more difficult for creators to remove these watermarks from generated images.
The invisible watermark feature is designed to be imperceptible to the human eye and will be applied using deep learning models within Meta AI. Although invisible, these watermarks can be detected using corresponding models. Meta asserts that its AI-generated watermarks can adapt to common image alterations such as cropping, color changes, and screenshots, making them more resilient compared to traditional watermarks.
Initially, this watermarking feature will be deployed within Meta AI, but the company intends to extend it to other Meta services leveraging AI to create images. Additionally, Meta AI's latest update includes "Reimagine" features for Facebook Messenger and Instagram, allowing users to send and receive AI-generated images. Consequently, both platforms will also incorporate the invisible watermark technology to enhance authenticity and track the origins of these AI-generated images.
The move by Meta comes amidst concerns about the misuse of generative AI tools, which have been utilized to create deceptive content, including fake videos and images of notable personalities, leading to potential scams and misinformation. Instances like an AI-generated image depicting an explosion near the Pentagon caused brief market disturbance, illustrating the consequences of such false information circulating online.
Moreover, human rights organizations, like Amnesty International, have encountered issues when inadvertently using AI-generated images in campaigns. Erika Guevara, Amnesty International's Americas Director, highlighted the organization's decision to remove such images from social media, emphasizing their commitment to focusing on supporting victims and seeking justice, steering away from controversies associated with AI-generated content.
