Meta, the parent company of Facebook and Instagram, has announced the addition of an invisible watermark to content created through its AI-powered chatbot. This step is part of their ongoing efforts to enhance AI tools and promote safety and transparency.
The company explained that many images created using their tools benefit from AI techniques to differentiate between human-generated content and AI-generated content. In the coming weeks, Meta will include an invisible watermark within the Imagine with Meta AI experience to enhance transparency and traceability.
The company stated that the invisible watermark will be added using deep learning models, remaining undetectable to the naked eye but can be revealed by a counter model. They also mentioned their flexibility in adapting to common image alterations like cropping and color changes without affecting the watermark.
This initiative follows requests from technology companies like Meta, Google, and Microsoft, operating in AI, to adhere to ethical principles in developing and implementing technology responsibly. Google previously introduced a tool called Synthoid to watermark images created by AI, similar to Meta’s approach of digitally embedding an invisible watermark directly into images, detectable by a specific model.
Ultimately, Meta announced the expansion of Imagine, a feature that converts text into images using smart technology, now available in the United States. This allows enthusiasts to create innovative images online for free.
