It's getting harder to spot counterfeits
New artificial intelligence (AI) technologies are becoming increasingly sophisticated and accessible, allowing almost anyone to create realistic images and videos. Gone are the days when we could confidently say, “seeing is believing.”
Unfortunately, this also means that scammers can use AI images and videos to make fake celebrity endorsements seem more authentic, or to trick families into thinking they're experiencing a fake emergency.
AI-Generated Text
If you're reading something and wondering if it was written by a human or generated by AI, consider these tips:
- Beware of typos: Typos are usually a red flag for scams, but if you're trying to distinguish between human and AI-generated content, typos can indicate that the content was written by a human.
- Beware of inaccurate or outdated information: Many AI bots are trained on limited data sets and may write outdated information.
- Are there repeated words or phrases? When discussing a single topic, the AI may use the same keywords and phrases it learned during training, which will sound unnatural.
- Pay attention to tone: AI writing can have a robotic, emotionless tone, whereas human writing can be more casual and use informal slang.
- When in doubt, seek out another source: If you are unsure whether the information you are reading is coming from a trustworthy source, seek out the information from another source.
AI-generated images
Here are some tips for recognizing AI-generated images:
- Check the context for red flags. Generative AI capabilities are advancing at an incredible rate, making it very hard to spot a fake. The best reason to be suspicious is the context in which you saw an image or heard a voice that may be fake. Ask yourself: Is the image or voice being used to pressure you to take a suspicious and urgent action, like sending money to a stranger or using a strange, unexpected payment method? If the image or voice is of a friend or loved one, don't respond until you've contacted that person to verify it's real.
- Find high-resolution images and zoom in. If you have doubts about a photo, look closely at the image in the highest resolution possible, search for it, and zoom in. There may be inconsistencies when looking at AI-generated images, which may provide some clues.
- Asymmetries and other physical inconsistencies should be considered, but with care. Less sophisticated AI may not handle physical features or unnatural body proportions well. Asymmetries in human faces, teeth, and hands are common in low-quality AI images. As AI technology improves, these mistakes will become less common.
- Weird textures and shiny effects. AI images often have textured backgrounds or an airbrushed look that isn't present in real photos. You may also notice strange backgrounds or clear images with random blurry areas.
AI-Generated Video
How to identify AI-generated videos:
- Watch out for strange shadows, blurring, and flickering lights. In some AI-generated videos, you may see shadows or flickering lights on just the face of the person speaking, or just the background. This is especially true in low-quality deepfake videos.
- Unnatural body language. Another telltale sign of AI. As with photography, it's difficult for AI to replicate a real person. If the body language seems off, you should question the authenticity of the video. Pay attention to movements and eye blinks. Listen closely; the synthesized voice might sound a bit odd. Choppy sentences, confusing intonations, and background sounds that don't match the speaker's location are all telltale signs of AI.
Bobby Hansen is regional director for the Better Business Bureau's Cedar Rapids office.
