Can You Spot AI-Generated Images? Tips for Identifying What’s Real Online

By Brynn Doering, Patron Services Librarian

With the growing use of artificial intelligence (AI), it’s becoming increasingly difficult to determine whether something online is real. Online content, such as text, images and videos, can be generated in seconds with the use of AI. Some of this content is convincing enough to trick even experienced internet users. After spending the last few years reading about AI, I've learned to spot many of the common warning signs, but even I still get tricked sometimes. Here are some tips that may help you identify questionable images online. 

The most important clues involve spotting flaws or nonsensical elements. Let’s imagine you’re viewing a picture of a person posing with their dog at a park. Look at the human subject first. AI is sometimes confused about anatomy and may produce an extra finger or merge body parts. Teeth, skin and hair may look airbrushed, creating an unnaturally perfect effect. On the other hand, the dog may have an extra ear, tail or leg. Let’s also say there’s a sign in the park that lists the rules for visitors. Are the letters strange swirls and shapes? Is the overall writing a bunch of gibberish? If the text is unreadable or impossible to decipher, the image may have been generated by AI. 

Keep in mind, however, that unusual details do not automatically mean an image is AI-generated. Camera angles, image compression, editing software or simple coincidences can sometimes create odd-looking results in genuine photos. It's best to look for multiple warning signs rather than relying on a single clue. 

Fortunately, we do have tools at our disposal for identifying AI-generated images. Some AI companies embed digital watermarks into images they generate. Tools such as Google's SynthID may be able to detect these markers, but not every AI-generated image contains a watermark and not every watermark can be reliably detected. There are also websites that attempt to identify whether an image was generated by AI by analyzing patterns within the image. AI detection tools can be helpful, but they’re not perfect. Different tools may produce different results, and both false positives and false negatives can occur. Before deciding whether an image is real, consider where it came from. Was it posted by a reputable news outlet, government agency or well-known organization? Is the same image being reported by multiple trusted sources? Reverse image search tools can also help determine whether an image has appeared elsewhere online or been altered from an original version. 

As AI tools continue to improve, identifying AI-generated images will likely become more difficult. No single method can reliably determine whether every image is real or artificial. When in doubt, use multiple strategies, look for unusual details, verify the source and check whether trusted organizations are reporting the same information.