How to Tell If a Picture or Video Was Made by AI

Good AI tools can now make a photo of something that never happened, or a video of a real person saying words they never said, in a few seconds. You run into this content every day in games, video apps and group chats, and they usually see it before any adult does. The goal isn't to make children suspicious of everything on the internet. It's to give them a habit, like the ones I learned in the military: keep your head on a swivel, stay aware of your surroundings, keep your back to the wall, and never take the same way home in case someone is watching. So when something online is surprising, upsetting or too good to be true, slow down and check before believing it or sharing it.

Step 1: Look for a label

Many platforms now mark AI-made or AI-edited content. Depending on the app, you may see a tag such as "AI info," "Made with AI," "AI-generated," or a note that content is "altered or synthetic." On YouTube, for example, creators are required to disclose realistic altered or synthetic content, and YouTube adds a label to those videos. In fact, I used AI to help edit this article.

A label is a strong clue that something was made or changed with AI. But no label does not mean it's real. Labels depend on creators being honest or on the platform's detection catching it, and both can miss.

Step 2: Check the Content Credentials

Content Credentials are a kind of digital "label" built on an open standard called C2PA. When a tool that supports it creates or edits an image, it can attach information about which tool made the file and when. OpenAI, for example, adds C2PA information and an invisible watermark to images it generates (OpenAI Help Center).

• Some apps and websites show a small "CR" icon on images that carry Content Credentials. Click or tap it to see the details.
• You can also upload a saved image to the free Content Credentials Verify tool from the Content Authenticity Initiative.

One important limit is that this information can be stripped out. Screenshots, reposts and file conversions often remove it, so a missing credential tells you very little. A credential that is there and names an AI tool tells you a lot.

Step 3: Ask Gemini whether Google AI made it

Google's Gemini app can check files for SynthID, the invisible watermark Google puts in content made with its own AI tools. Upload the image, video or audio at gemini.google.com and ask, "Was this created or edited by Google AI?" According to Google's help page, the watermark check only recognizes Google AI content, videos need to be under 90 seconds, and there's a daily limit on checks. Gemini also reads Content Credentials from other companies' tools when they're present. A "no" from Gemini means "not made with Google AI." It doesn't mean "definitely real."

Step 4: Search for where the image came from

A reverse image search is one of the most useful checks, and kids can learn it quickly. Use Google Lens or TinEye to upload the picture or a screenshot of it. You're looking for:

• The earliest version. Did it first appear on a news site, a photographer's page, or an account known for AI art?
• Other versions. Has the same scene been posted with different captions, dates or details?
• Fact-checks. If an image is going viral, a news organization or fact-checker may have already looked into it.

For videos, search for the event itself. If something dramatic really happened, reliable news outlets will usually be reporting it too. If only one account has it, be careful.

Step 5: Ask who posted it and why

Much of the time the most revealing question isn't about pixels at all. Who shared this? Is the account new, anonymous, or built around getting clicks? Is the post trying to make you angry, scared or excited enough to share it right away? Strong emotions are exactly what fake content is designed to create, so a strong reaction is a signal to slow down.

Step 6: Look closely, but don't count on glitches

AI images and videos still sometimes give themselves away. Things worth checking:

• Hands, teeth and ears with the wrong number of parts or odd shapes.
• Text on signs, shirts or books that turns into nonsense up close.
• Backgrounds where lines, railings or faces in the crowd melt together.
• Lighting and shadows that point in different directions.
• In videos: mouths that don't quite match the words, unnatural blinking, and edges around the face or hair that shimmer.

These clues are getting rarer as the tools improve, which is why they come last on this list. Teach kids that finding a glitch is a good sign something is fake, but not finding one doesn't prove it's real.

A simple routine for kids: Pause, Check, Ask

For younger kids, three words are easier to remember than six steps:

1. Pause. If a picture or video makes you feel something strongly, don't share it yet.
2. Check. Is there a label? Is anyone else, like a news site, showing the same thing?
3. Ask. Show a parent or teacher and figure it out together.

It also helps to talk about why this matters beyond "being fooled." Fake images can be used to embarrass or bully real people, including classmates. A good rule for kids is simple: never make, share or forward an AI image of a real person without their permission, and tell an adult if someone does it to you or a friend.