How Can We Tell If Videos and Images Are AI-Generated?

Any tips that work today might not apply tomorrow, with so many AI models showing different behaviors! A while back, counting fingers was a thing to look out for, but AI has improved so much that it's not a reliable method anymore. It’s crucial for everyone to become media literate—question everything, verify through multiple credible sources, and stay skeptical if something seems too sensational.

Honestly, it's getting harder to tell what's real and what's not. Some might say we don't even need to worry, since a lot of the stuff we see – like on TV – isn't real anyway, it's all just acting. But if you want to find clues, one thing I've noticed is that AI videos often have outputs with resolutions that can be divisible by 24. Of course, you can crop a video to disguise that, so it’s not foolproof.

Keeping up with the tech will really help you notice changes over time.

It's shocking how easily audio can be manipulated with AI, making it seem like people said things they never did. This raises huge questions for politics and law enforcement. Video evidence, which is usually trusted, is becoming questionable. I wonder if police will even have the capability to check if footage has been altered.

With how quickly AI is evolving, I honestly think that by next year, distinguishing between AI and real will be nearly impossible. Just look at how much better the AI-generated stuff has gotten in just a year! Even if they introduce laws to watermark AI outputs, it still won't guarantee authenticity. The whole situation feels pretty hopeless.