Artificial Intelligence is at the change of how we create and present media. We see this in tools like Google Veo, Sora by OpenAI, and the many AI Generated Images which are now at your disposal to produce very real looking fake images and videos. While we benefit from these great creative tools we also must address the issues they bring — in particular the spread of false info.
In the past few months we have seen a large scale adoption which we may term as addiction of real looking but false content in Pakistan and world over. From an enormous deep faked image of a goat at the Eid cattle market going viral to that of a celeb saying out of the blue offensive things — we are put to the test in terms of which info we can still trust.
In our report we will guide you through the process of identifying which images and videos are AI generated we also cover practical tools, common sense tips, and a critical approach.
Why is it That We are Finding it Hard to Identify AI Generated Content?
AI development has seen great growth. In the field of video and image production which includes that of Veo and Mid Journey we see very real results — almost photo or life like. At first you won’t be able to tell the difference between what is real and what is generated. And that’s the problem.
What in the past was the domain of special effects studios is now available on mobile phones which in turn has made it easy for anyone to spread disinformation, fake news, or manipulated stories.
But the good news is: AI still leaves signs — which you can learn to recognize.
1. Always Look at the Source.
As you encounter shocking, heartwarming, or unbelievable reports put out there — stop to question their source.
- Did it come from a verified account?
- Does it come from a reliable news source?
- Is the original source mentioned?
If not or if the material is from a new and unproven source which may have little to no reputation — be careful. Fake info often appears in non official channels or in short term social accounts.
Tip: Check out sites like Who Posted What and also go through an account’s history to see if they post mostly real or fake info.
2. Lack of Info — Red Flag.
AI generated images and in many cases videos appear out of the blue — they lack:
- Date or time of the event
- Location where it occurred
- Verified eyewitnesses or sources
If media is silent on a story or what they put out is inconsistent upon fact check — that is a red flag. Go out and find other articles, tweets or news reports. Also if no main platform is reporting the same event, it may not have transpired at all.
3. Look at the Details.
Even at the top of their game AI tools will make mistakes. When you’re skeptical of an image or video check in with a closer look. Here is a list of which common AI created flaws to look out for:
- Distorted fingers, ears, or facial expressions
- Inconsistent lighting or strange shadows
- Asymmetrical clothing or backgrounds
- Misplaced text on signs, boards or banners
- Missing reflections or bizarre textures
At first you may not see these small mistakes, but once you know what to look for they stand out.
Example: If you see a photo of a politician at a protest which may be doctored check the details of the hands, face structure, also the background — AI frequently messes up there.
4. Use of Reverse Image and Video Tools.
Reverse engineering tools are very easy to use:
- Google Reverse Image Search: Upload an image or paste a link to see where it appears elsewhere.
- TinEye: Helps you track down the primary or first version of an image.
- InVID: A fantastic resource for reverse searching videos, particularly clips which are shared on social media.
- Amnesty YouTube DataViewer: Lets see what other uses there are for video thumbnails.
If you notice the same image is present in different events or timelines that is a great indicator it may be fake or AI generated.
5. Metadata Uncovers the Truth.
In most cases which includes images and video we see metadata which is info like the date of creation, editing tools, and the device that was used for the content.
Use EXIF Reader, FotoForensics, or Jeffrey’s Metadata Viewer.
AI-generated content often shows signs like: Created in MidJourney, DALL-E, or edited with Runway ML.
Although we may try to remove it, in most cases metadata remains in which we don’t expect it to and that which confirms if an image was taken by a real camera or produced via a computer.
6. Trust Your Instinct But Verify with Tools.
If a situation seems too good to be true, too shocking, or far out of the ordinary — trust your instincts but also support them with data.
Ask yourself: Reflect upon it:
- Does the content seem exaggerated?
- Is it emotionally manipulative?
- Would a big name media player look past it is that were real?
Many people are fooled by political posts which are put out by AI, fake medical information, or passed off as real celebrity content. Don’t react first, instead take a step back to check your info out and decide.
7. Follow and Engage with Fact Checkers and Also Learn More.
As technology grows in complexity it is to your benefit to stay at the forefront of change. Support yourself with trusted fact checkers and digital literacy groups:
- AFP Fact Check
- Snopes
- Pak Check
- Poynter’s MediaWise
They report on what is trending in fake news and how to identify it.
What is the set of AI tools which are used for fake content?
- Midjourney, DALL-E — for image generation.
- Sora, Veo, Runaway — for video generation.
- FaceSwap, DeepFaceLab — for making deepfakes.
- ChatGPT and Image Models — in use for fake news articles which include images.
- AI Voice Synthesis — to reproduce real voices in video.
If there are signs of any of these issues do an in depth look.
Conclusion
AI generated content is a present day reality which we see — it’s a fact of life. While the technology itself is not in and of itself bad, it is put to bad use at times for the spread of misinformation, scams, and manipulated stories.
The best defense is awareness.
By way of recognizing what is true and what is not in AI generated media and by fact checking what you are going to share, you may play a role in reducing the spread of fake content which in turn protects others from misinformation.
