Misinformation Education Creative Activity: AI misinformation/deepfakes and how to spot them

Blog Post #5

For our final blog post, we had to choose to create a credible message about misinformation, and in a creative way. Being that I really enjoy blogging, I chose to do a blog, but with a little video element mixed in along with the usual links and images. I also chose to do it on AI, and if you read my other blogs you would know that AI has been a fixture in almost all of them. This includes an example of AI that fooled me, and how you can use the SIFT method to deduce if something is AI-generated. Anyways, before the blog gives you all the info you need, here is an introductory video AI.

In this video I go over a little bit about some of the dangers of AI. Articles I reference briefly as examples are a woman being scammed by a person impersonating Brad Pitt through AI, AI voice cloning that scams victims over the phone, and someone who gave $25 million to a scammer using deepfake technology over a video conference call. I also give you a little scammer of how simple it is to use some of this technology. The websites used for my example are Voice Booking and Magic Hour.

What is AI?

When you think of AI, your brain may go to famous examples like the villainous HAL-9000 in the film 2001: A Space Odyssey. But while AI hasn't progressed to the point that it is self-aware, it can be used for nefarious purposes by people to spread misinformation through AI-generated media.

AI, or artificial intelligence, refers to computational systems that can perform complex tasks similar to that of a human. This includes math, art, writing, and of course speaking. These programs learn by taking in data and learning patterns that allows it to mimic what it "sees." So while AI cannot create, it can learn and utilize that. According to NASA, AI is so vast in scope and ability that there "is no single, simple definition of artificial intelligence" (NASA).

Famous examples of widely-used AI include ChatGPT, which can do things like write entire essays for people, and Grok, which was developed by Elon Musk's xAI company and exists primarily on the X/Twitter platform as a chatbot (a software program designed to imitate human conversation) or an AI image generator.

Grok's logo.

What are Deepfakes?

Deepfakes are a type of AI-generated image, video or audio, usually meant to mimic specific individuals. For example, you can plug examples of someone's voice into a deepfake generator, and it can analyze it and make it possible to make them say effectively anything, sometimes incredibly convincingly. According to the National Council on Aging, deepfakes "can make it appear a person has said or done something they never did" (NCO, 2024). As far as making people look like they said or did something, AI can even make still images move in a lifelike way.

An example of deepfakes from Voicebot.ai making still images move in a lifelike way, including paintings. The example above is from Samsung Lab's tool "MegaPortraits."

Deepfakes have many different types and results. Some deepfakes are just putting a face over a person to make them look like someone else, as I did. Or Ai can create an entire person from scratch, which sometimes results in crazy images that look so outlandish that you wonder how people could ever fall for them. This includes a photo of the recently deceased Pope Francis dressed like a pope-styled rapper, including a puffer jacket, diamond-studded cross, and carrying what appears to be holy water. As I once fell for an image, a writer for the Guardian, Josh Golby, also did. No matter your age, Golby and I both found that AI can be convincing enough to fool anyone; "I thought I was above being hoodwinked in this way" (Golby, 2023).

An AI-generated image of Pope Francis, created by Midjourney. While some AI and deepfakes simply add features to a real image, this example shows how AI can create entire images by itself. Another article by Kelvin Chan and Ali Swenson of the AP goes over the ways you can deduce this image is AI.

How AI can be used for misinformation, and how to spot it

AI can both be very easy to spot, and very difficult. One of the big issues I have noticed is when AI is used during real-world events and put alongside real images and news. This can make it even more difficult for people to deduce what is real and what isn't as I explored when I analyzed an AI Tweet about a protest in Turkey that fooled me. In that article example, I used the S.I.F.T. method, or stop, investigate, find better sources, and trace the claim back to original context.

One of the many fake AI images post in relation to the Myanmar Earthquake that occurred on March 28, 2025. Images like these show how AI can contribute to the spread of misinformation about real-world events (retrieved from DW).

STOP

Using the SIFT method, in my opinion, is the best way a person can deduce if an image or video is AI-generated. To start, you actually STOP and look at the image. Does it look real? Is there any weird stuff going on with any of the people in the image? Maybe some words present are misspelled, or maybe someone has extra fingers. Simply stopping and looking at an image a little closer for any discrepancies in reality can deduce if it is AI-generated.

An AI image I generated via Grok of two people shaking hands. Looking closely, you can see that the person on the right has six fingers on the hand they are shaking with, and easy way to spot an AI image.

INVESTIGATE

If the image has no immediate discrepancies, a good way to see if the source of the image is reputable is to see see what else the account you saw it on has posted. Are they someone who posts a lot of political content? Do they put anything related to AI in their social media bios? A quick look into who someone is, or at least how they show themselves online, is a good way to deduce if it is AI misinformation. And just because this account posted does not mean that is where the image came from.

Example of someone posting a AI photo (likely didn't create it) and not signaling that it is AI in any way in the description or hashtags. As far as we know, they intend it to be taken as real, and even if they don't, someone still might.

FIND BETTER SOURCES

If investigating the account that posted it is not enough, then try to find the image elsewhere. Have any reputable news sources posted it? Or maybe check a misinformation debunking site such as Rumor Guard, which has up to date information on recent attempts at misinformation online including AI-generated posts. You may find what you are seeing on there, as I did with the AI that fooled me.

The current front page of Rumor Guard, which has a debunked example of AI misinformation. If something you are seeing is AI, there is a good chance Rumor Guard has looked into it.

TRACE BACK TO ORIGINAL CONTEXT

If the previous three steps do not get an answer for you, this last step should. If you can find the original account that posted it, which if you engaged in the previous three steps you likely found, you can see who posted it and why. Did they intend it to taken seriously? Are they themselves spreading misinformation on purpose? And if you cannot find the original source at all, no matter how much you search, then that is even more proof that the media you are seeing is likely not real.

Example of an account that posts AI content, Eren Fazlıoğlu, who posted content that was used by others as real and fooled me.

Conclusion on AI misinformation

AI misinformation is becoming a bigger issue every day. As it continues to get more and more advanced and easier to use, the amount of it online will only increase. Hopefully by reading the blog and seeing my video, you leave with a better understanding of what AI is, how ti can be used for misinformation, and how you can detect it. As mentioned in my video, we have already seen major examples of people not only being misinformed by AI content, but scammed out of large sums of money. Take the finance worker in China who was duped in a video conference call with multiple deep fakes to pay out $25 Million! The scary thing is that the worker had his doubts initially, but "put aside his early doubts after the video call because other people in attendance had looked and sounded just like colleagues he recognized" (Chen and Magramo, 2024).

Even informed skeptics can be quickly convinced by deepfakes, like this man and myself. But by applying the SIFT method and being more aware of this technology, you can better help yourself detect it, and stay safe from being fooled by it.

For some other examples on avoiding misinformation, read my other blogs!

Thank you.

References

Chan, K., & Swenson, A. (2024, March 21). One Tech Tip: How to spot AI-generated deepfake images | AP News. AP News. https://apnews.com/article/one-tech-tip-spotting-deepfakes-ai-8f7403c7e5a738488d74cf2326382d8c

Chen, H., & Magramo, K. (2024, February 4). Finance worker pays out $25 million after video call with deepfake ‘chief financial officer.’ CNN. https://www.cnn.com/2024/02/04/asia/deepfake-cfo-scam-hong-kong-intl-hnk

De Graaf, A., & Muros, C. (2025, April 10). Fact check: Fake news on Myanmar, Thailand earthquake. dw.com. https://www.dw.com/en/fact-check-fake-news-on-myanmar-thailand-earthquake/a-72103584

Garris, T. (2025, May 3). AI voice-cloning scams target victims with deepfake technology. WMUR. https://www.wmur.com/article/ai-voice-cloning-scams-deepfake-5225/64646191

Golby, J. (2023, March 27). I thought I was immune to being fooled online. Then I saw the pope in a coat. The Guardian. https://www.theguardian.com/commentisfree/2023/mar/27/pope-coat-ai-image-baby-boomers

Gozzi, L. (2025, January 15). French woman duped by AI Brad Pitt faces mockery online. BBC. https://www.bbc.com/news/articles/ckgnz8rw1xgo

How to spot Deepfake Scams. (2024, October 30). NCO. https://www.ncoa.org/article/understanding-deepfakes-what-older-adults-need-to-know/

Magic Hour. (2023). Magic Hour AI Face Swap (May 2025 version) [Generative AI]. https://cset.georgetown.edu/article/what-are-generative-ai-large-language-models-and-foundation-models/  

Matthias, M. (n.d.). Why does AI art screw up hands and fingers? Britannica. https://www.britannica.com/topic/Why-does-AI-art-screw-up-hands-and-fingers-2230501

No, a creature called ‘Sleeping Rose Mantis’ isn’t real. (2025). https://www.rumorguard.org/post/no-a-creature-called-sleeping-rose-mantis-isn-t-real

Schwartz, E. H. (2022, September 19). Samsung’s New MegaPortrait Generates Deepfake Videos from Still Images - Voicebot.ai. Voicebot.ai. https://voicebot.ai/2022/08/01/samsungs-new-megaportrait-generates-deepfake-videos-from-still-images/

Voicebooking. (2011). Voicebooking AI Voice Generator (May 2025 version) [Generative AI]. https://www.voicebooking.com/en/free-voice-over-generator 

Viral image of Pikachu at protest is AI-generated. (2025). https://www.rumorguard.org/post/viral-image-of-pikachu-at-protest-is-ai-generated

What is Artificial Intelligence? - NASA. (n.d.). NASA. https://www.nasa.gov/what-is-artificial-intelligence/

xAI. (2023). Grok (May 2025 version) [Large Language Model]. https://x.ai/