While many have begun to question the legitimacy of images as AI generation nears photorealism, Google looks to be the first to challenge this. Last week, Google announced, through its partnership with Shutterstock and Midjourney, that it would classify any images that were AI-generated.
Google’s new solution will look to provide context regarding the origins of the image, including where it first appeared. This solution will also look to clarify where else across the internet an image may have appeared. Finally, any proof from any news organisations which identify the simulated nature of the image will be referenced.
This is a significant move by Google, as it is one of the first steps to reduce misinformation in the AI space. One of the biggest issues with AI images is their use in creating fake truths, where realistic imagery can convince people of an unlikely situation. Or worse, realistic ‘photos’ of a believable situation. Given Google’s dominance in the search space, their attempt to help classify images across their search page helps to legitimise their results.
The most encouraging facet of this move by Google, is that they are doing this before any regulations are in place to force their hand. While the European Commission recently began making amendments to its AI bill, nothing in these changes seeks to mandate anything like Google’s product. As Google continues to build out its AI solution, they clearly have a grasp on the ethical concerns and are already beginning to address them.
Not only is this a big move by Google, but it is an encouraging move by Midjourney, who have developed the best solution for making convincing AI-generated images. Midjourney currently represents one of the biggest AI-generated platforms in the market. Additionally, Midjourney’s stable diffusion models are the most capable of producing photorealistic images.
Midjourney clearly has a focus on ensuring its platform is used to the highest standards and does not risk any potential misuse. Midjourney frequently looks to ban users who do not abide by its strict guidelines around the output. Midjourney has pulled its free version sporadically as a number of free users were abusing the platform.
This is an interesting move by Shutterstock, which does not allow AI-generated images to be used on their site. However, this is something that will benefit both Google and Shutterstock. Google will be able to leverage Shutterstock’s large library of content for reference. Meanwhile, Shutterstock can help to ensure that any images uploaded onto their site are in line with their brand guidelines.
While this will not solve some of the emerging issues with AI-generated images, it is an encouraging first step. Google is clearly engaged in the ethical questions surrounding this topic and has proactively started to take steps to combat it. The largest search engine in the work, partnering with one of the biggest AI-generated image platforms and image content libraries establishes a big step towards making the internet a safer place.