![meta](https://wikitechlibrary.com/wp-content/uploads/2023/12/meta-1.webp)
Meta Platforms announced this week that it will add invisible digital watermarking technology to images created by its AI text-to-image generator, Imagine. The feature will be rolled out over the next few weeks.
The watermarks are designed to be resilient to common image manipulations like cropping and screenshots.
Their purpose is to enhance transparency around AI-generated content as Meta continues to expand access to and uses for its generative AI products.
Imagine is part of Meta’s overarching Meta AI platform. It was launched in September 2022 alongside other AI offerings meant to showcase the company’s investments in artificial intelligence research.
Meta says Imagine’s watermarking feature is just the first of many similar transparency tools the company hopes to bring to its AI product lineup going forward.
The announcement comes on the heels of rising concern over the potential for misuse of generative AI models like text-to-image generators.
ChatGPT’s viral popularity in recent weeks has drawn fresh scrutiny to AI’s societal impacts.
By preemptively adding watermarking before expanding Imagine access, Meta aims get ahead of ethical debates about attribution and misrepresentation of AI-generated content. The company says its goal is to balance innovation with responsible AI development.