Future of AI-Generated Images – What Are the Cyber Risks?

Nowadays, there are plenty of pictures generated by AI. It is often used to create images of real people, and these images are of such high quality that it can sometimes be difficult to distinguish between AI-generated images and real ones. While there are often some obvious signs to differentiate AI-generated images, in some cases, the artifacts are either minimal or barely visible.

In this article, we will discuss the risks associated with AI-generated images. We will also provide examples of what these images look like and how to detect them.

Generating an Image with AI – Case Study

To illustrate AI-generated people, we used the virtual image generation software Genfluencer.ai.

The first step in using the platform is to generate a face. The following image was generated by specifying just a few parameters, such as gender, age, hairstyle, and color:

Creating a face first ensures that other images will feature the same, consistent person. While a face can be created with just a few features, a well-crafted prompt is necessary to produce realistic images.

I decided to use AI to generate a prompt. Who communicates with AI the best? It is the AI itself! I asked ChatGPT to provide a “super realistic” image of a woman, specifying that I would use a face image as input, so there was no need to describe facial features.

This is the prompt that ChatGPT provided:

A super realistic, full-body image of a young woman in stylish, high-quality clothing. She stands confidently in an elegant outdoor setting, with a well-designed background featuring modern architecture or a scenic landscape. The lighting is natural and flattering, highlighting the texture of her outfit and the richness of the environment, creating a polished and sophisticated look.

I used this prompt, along with the previously generated face, to create an image using the Genfluencer.ai image generator. This is the image I received:

This outcome is a super realistic. And there are no obvious signs to detect that this is a fake person.

So, What Are the Cyber Risks of AI-Generated People?

There are several ways in which AI-generated images can be used for malicious purposes, including:

• Misinformation: one of the most significant dangers of AI image generation is the creation of deepfakes. These highly realistic images or videos can impersonate real people, such as politicians, to spread false information or incite panic among the public. Deepfakes can easily deceive viewers, making them a powerful tool for disinformation.
• Scams: scammers are always on the lookout for new ways to exploit people, and AI-generated images have become a potent tool in their arsenal. By creating realistic but fake profiles, scammers can impersonate others, tricking individuals into forming emotional connections or sending money. The realistic nature of these AI-generated personas makes these scams even more convincing and dangerous.

How to Spot Images Generated with AI?

Spotting AI-generated images can be challenging, especially as the technology becomes more sophisticated. Nowadays, not only the images can be generated, Open AI is creating a tool that can be used for generating super realistic videos. However, there are several techniques and signs that can help identify such images:

• Unnatural Hands and Teeth– AI struggles with generating human hands and fingers. Look closely, and you might notice that the hands have more fingers than a real person. AI also often produces distorted teeth.
• Weird Backgrounds: Take a closer look at the background, and you may see strange, unrealistic objects or people that appear unnatural.
• Clothing and Accessories: Check for inconsistent or nonsensical patterns in clothing, like stripes that don’t align or strange fabric textures.
• Facial Features: Mismatched eyes or ears, irregularly shaped pupils, or eyes that aren’t aligned correctly are different signs that can indicate the image is not real.