Artificial intelligence can now produce photographs of people who do not exist and videos of real people saying things they never said. As the quality improves, the old advice to simply look for obvious mistakes is no longer enough. Still, a mix of visual checks and good habits can help you judge whether an image or video deserves your trust before you react or share it.
Visual clues that still slip through
Generated images have improved dramatically, but they often stumble on fine, repetitive detail. Human hands remain a common weakness, sometimes showing the wrong number of fingers or unnatural bends. Look closely at areas where the model has to keep track of many small elements, because these are where errors tend to appear.
- Hands, teeth, and ears that look subtly wrong or asymmetric.
- Text on signs or clothing that turns into meaningless shapes.
- Jewelry, glasses, or patterns that blend into skin or change midway.
- Backgrounds where objects merge, warp, or repeat oddly.
- Lighting and shadows that fall in inconsistent directions.
In video, watch for faces that flicker at the edges, mouths that do not quite match the words, and blinking that seems too regular or absent. Hair and the boundary between a face and its background are hard to render smoothly, so glitches often show there.
Why looking is not enough
The uncomfortable truth is that the best fakes now pass casual inspection, and the tell tale flaws are disappearing with each new model. Relying only on your eyes will eventually fail. That is why the more durable skill is not spotting pixels but questioning sources. A convincing image with no credible origin should raise more suspicion than a slightly flawed one from a trusted outlet.
Habits that protect you
Treat surprising or emotional media the way a careful reader treats a surprising headline. The goal is to slow down before you believe or share.
- Ask where it came from and whether a reputable source also reports it.
- Search for the same event from independent outlets or official accounts.
- Use a reverse image search to see where the picture first appeared.
- Check the date, since old images are often recycled as if they are new.
- Notice if the content is designed to make you angry or afraid, a common tactic.
Context is often the strongest signal. A dramatic clip that only appears on anonymous accounts, with no coverage from established newsrooms, is a warning sign regardless of how real it looks. Genuine major events tend to be documented from many angles by many people.
The role of labels and detection tools
The technology industry is developing ways to mark and trace synthetic media. Some systems embed invisible signals or attach a record of how an image was made, and camera makers and platforms are beginning to support these standards. Automated detectors also exist, though they are imperfect and can be fooled. These tools are helpful as one input, but they should support your judgment rather than replace it, since no single check is reliable on its own. It also helps to remember that provenance signals confirm where genuine content came from more than they prove that unlabeled content is fake, so their absence is not by itself evidence of a forgery.
Finally, be aware of your own biases. We are quicker to accept media that confirms what we already believe and quicker to doubt what challenges it. A fake that flatters your existing opinion can slip past your guard precisely because you want it to be true, so apply the same scrutiny to content you agree with as to content you do not.
The bottom line
You can still catch many fakes by studying hands, text, and edges, but the surer defense is skepticism about sources. Slow down with anything shocking, confirm it through independent and reputable outlets, and lean on emerging labels and tools as one clue among several. In a world of convincing synthetic media, the most valuable skill is not spotting flaws but verifying before you believe.