It’s International Fact-Checking Day. Refresh your AI identification skills

AI-generated content is everywhere these days, making it increasingly difficult to separate fact from fiction, especially when it comes to breaking news.

Look no further than the war in Iran. Since the United States and Israel attacked Iran on February 28, researchers have identified an unprecedented number of false and misleading images generated by artificial intelligence that have reached countless people around the world. Among them are fake images of bombings that never happened, images of soldiers who were allegedly captured, and propaganda videos created by Iran that depict President Donald Trump and others as blocky, Lego-like miniatures.

Thursday, the 10th annual International Fact-Checking Day, provides a good opportunity to examine these evolving challenges.

Disinformation created through AI is being shared at unprecedented speed from an infinite number of sources. From the start of the war in Iran, narratives from all sides in the conflict encouraged this type of content.

The Institute for Strategic Dialogue, which tracks disinformation and extremism online, examined social media posts regarding the Iran war. Among their findings was a group of X accounts that regularly post AI-generated content and have collectively gained more than a billion views since the conflict began. This was done by about two dozen accounts, many of which had blue check verification.

Here are some tips for distinguishing AI-generated content from reality in an online world where this is becoming increasingly difficult.

When AI-generated images began to spread widely online, there were often obvious clues to identify them as fabricated. Maybe a person had too few – or too many – fingers, or their voice was out of sync with their mouth. The text was perhaps absurd. Objects were frequently distorted or missing key components. As technology continues to evolve, these clues are no longer as common as they once were, but they are still worth looking for. Watch for inconsistencies, like a car appearing in a video at one moment and disappearing the next moment or actions that are not possible according to the laws of physics. Some images may also be too polished or have an unnatural sheen.

AI-generated images are shared again and again. One way to determine their authenticity (or lack thereof) is to research their origin. Using a reverse image search is an easy way to do this. If you’re watching a video, take a screenshot first. This could lead to a social media account that specifically generates AI content, a distorted older image, or something completely unexpected.

Search for several verified sources that can help authenticate the image. For example, this could mean a fact check by a reputable media outlet, a statement from a public figure, or a social media post from a misinformation expert. These sources may have more advanced techniques for identifying AI-generated content or accessing image information that is not available to the general public.

There are many AI detection tools that can be a useful starting point. But be careful, because their assessments are not always correct. Images generated or edited with AI using Google’s Gemini app include an invisible digital watermarking tool called SynthID, which the app can detect. Other AI creation tools have added visible watermarks to the content they generate. However, they are often easy to remove, meaning that the absence of such a watermark is not proof that an image is authentic.

Sometimes it’s just about getting back to basics. Stop, breathe, and don’t immediately share something you don’t know is real. Bad actors often rely on people letting their existing emotions and viewpoints guide their reactions to content. Reviewing comments can provide clues as to whether or not the image you are looking at is real. Another user may have noticed something you didn’t or was able to find the original source. Ultimately, it’s not always possible to determine with 100% accuracy whether an image is AI-generated, so be mindful of the possibility that it’s not real.

See something that seems false or misleading? Email us at FactCheck@ap.org.

___

Find AP Fact Checks here: https://apnews.com/APFactCheck.