Snapchat will add a watermark to AI-generated images

MADRID, (Portaltic/EP) – Snapchat announced that it will soon add a watermark to all images on the platform that have been created by generative Artificial Intelligence (AI) and will do so when these creations are exported or saved on the camera roll.

The company believes that platform users “should be informed about the types of technologies they use”, for which the firm already uses icons, symbols and labels that provide context and information when they interact in text-based conversations with the ‘ chatbtot’ My AI, powered by ChatGPT.

It also said it is “very careful” about vetting political ads through a human review process, including a “thorough” check for any misleading use of content or generative AI.

To provide more context about what’s being shared, it announced that it will soon add a watermark to images created with Snapchat’s AI tools. This will be seen when the image is exported or saved to the camera roll.

In this way, the recipients of these images generated with AI will be able to see the Snapchat logo with a star in the upper right, which indicates that a tool powered by this technology has been used to create said content.

Snapchat updated its Help Center to advise users not to remove this watermark, as it violates the platform’s Terms of Service. Likewise, he has advised not to assume that the results of AI are true, because it is a technology “intended to alter or transform an image.”

COMPLY WITH PRIVACY REQUIREMENTS

Snapchat clarified that all AI-powered features always undergo strict review to ensure they meet its security and privacy principles. For this reason, it has the so-called AI red team, which tests the consistency of AI results and identifies flaws in the models to implement security solutions.

It also recalled that it has a series of responsible governance principles and that it has created a security review process to detect and eliminate potentially problematic messages in the early stages of development of AI Lens experiences. Likewise, he has said that, taking all this into account, he is implementing additional testing to minimize potentially biased AI results.