Google researchers have sounded the alarm on an explosive rise in AI deep fake images, warning that these hyper-realistic fakes are spreading fast. This surge in digital deception is not only blurring the lines between reality and fiction but also posing significant threats to information integrity, trust, and security in our increasingly digital world.
The findings reveal the rapid adoption of this technology by those intent on spreading false information. However, researchers cautioned that AI-generated images are merely one method of deception — the most prevalent remains the misuse of genuine photos taken out of context.
Researchers released an online paper this month, not yet peer-reviewed, that analyzed nearly 136,000 fact-checks spanning from 1995 to November 2023. Their findings showed that AI-generated image misinformation was negligible until spring 2023, around the time fake photos of Pope Francis in a puffer coat went viral. Notably, most of these fact-checks were published after 2016.
The study found that approximately 80% of fact-checked misinformation claims involve media such as images and videos, with video increasingly dominating these claims since 2022. Despite the rise of AI, real images paired with false claims about their content or implications continue to spread widely, often without the need for AI or any photo-editing.
Although the study indicated that AI models typically aren’t trained to generate images such as screenshots and memes, it is likely they will soon learn to produce these types of images as advanced language models evolve.
Read the complete news article on NBC News