Detect AI Generated Images: Manager's Verification Checklist

Why You Need to Care About Fake Images Right Now

Your competitor just posted a beautiful before-and-after case study on LinkedIn. The lighting is perfect. The client testimonial is glowing. The ROI numbers look impossible to ignore.

It might be completely made up.

According to recent data, approximately 35% of B2B marketing teams admit to experimenting with AI-generated promotional imagery. That's not even counting the teams that are doing it without admitting it. For you as a manager, this creates a real problem: fake images are influencing real business decisions. Your team might be trying to match competitor performance metrics that don't actually exist. You might lose a hire opportunity because a competitor's "culture photos" look better than your real ones. Or worse, you might waste budget trying to replicate a case study that was assembled in Midjourney rather than shot with an actual client.

The good news? Detecting AI-generated images is getting easier once you know what to look for. You don't need AI detective software. You just need to look carefully and ask the right questions.

The Tell-Tale Signs: What AI Images Get Wrong

AI image generators have gotten scary good. But they still have predictable weak spots. Learning these patterns is like building an eye for spotting bad Photoshop work, except now you're catching Midjourney, DALL-E 3, and Gemini 2 instead.

Hands and Fingers Are Still a Disaster

This is your easiest win. AI generators struggle with human hands like your uncle struggles with video calls. Look for extra fingers, missing fingers, fingers bent at impossible angles, or hands that look like they're melting into the arms. Check both visible hands carefully. A CEO in a stock photo should have five distinct, believable fingers on each hand.

Real example: You're evaluating a competitor's "team collaboration" image on their website. Four employees are sitting around a table reviewing documents. Look at their hands on the table. In a real photo, you'll see clear joint definition, realistic nail appearance, and natural finger positioning. In an AI image, you might see six fingers on one hand, or fingers that seem to stretch too long, or hands that look slightly translucent where they overlap objects.

Eyes and Faces Have Weird Symmetry Issues

Human faces are asymmetrical in specific ways. AI often makes them *too* symmetrical or symmetrical in the wrong places. Look at the iris size (are they identical?), pupil position, and eye depth. One eye might be slightly larger or at a different angle than the other. Real faces have this. AI faces sometimes don't.

Also check eyebrows, eyelashes, and eye shine. AI is improving here, but you'll sometimes see eyelashes that are too perfect, too uniform, or positioned strangely. The catch light in the eyes (that little white reflection) might be identical in both eyes instead of naturally slightly different.

Text and Numbers Look Normal But Read Wrong

AI has made huge progress on text generation, but visible text in images still trips it up regularly. Look for misspelled words, backwards letters, distorted numbers, or text that's grammatically weird. Sometimes the text will be there but slightly blurry or pixelated in an unnatural way.

Real example: You're reviewing a competitor's case study image showing performance metrics. The chart displays a line graph with axis labels. In an AI image, those labels might spell out something like "Converson Rate" or show "$1,2450" with inconsistent decimal placement. Or numbers might reverse themselves (a "9" might be upside down or sideways). Real screenshots have clean, correct text. When you see text anomalies, that's a red flag.

Objects Blend or Distort at the Edges

AI generators struggle with object boundaries, especially where items overlap or meet backgrounds. Look for blurry transitions, objects that seem to merge into one another, or items with soft edges that should be sharp (like a laptop keyboard or business card). Background elements sometimes blur weirdly or fade unnaturally into the foreground.

The Verification Checklist: Your Step-by-Step Process

You don't need to be an AI expert. You just need a system. Here's one that works.

1. Zoom in on hands first. Open the image at full size. Check every visible hand. Count fingers. Look for distortion. This catches 40% of obvious AI images immediately.
2. Scan for visible text. If there's any text in the image (signage, papers, devices), read it carefully. Is it spelled correctly? Do numbers make sense? Is the text too blurry or distorted?
3. Study the eyes and face symmetry. If there are people, zoom in on their faces. Do eyes look naturally proportioned? Is the symmetry too perfect? Are pupils positioned naturally?
4. Check object boundaries. Where do objects meet the background? Are there weird blurs, halos, or soft edges that don't make sense for that object type?
5. Look for impossible physics. Does the lighting make sense? Are shadows consistent? Do reflections match the light source? AI sometimes creates physically impossible scenarios.
6. Reverse image search. Use Google Images or TinEye. If an image appears in dozens of unrelated articles or websites, it's probably a stock photo. If it appears nowhere else, that's interesting. If it appears only on competitor websites claiming it's their own work, that's a signal.
7. Ask for the original file. Real professional photography exists in full resolution. If a competitor can't provide source files or a photographer credit, ask them directly. This isn't accusatory, it's just business due diligence.

This checklist takes 90 seconds per image once you get practiced. Build it into your competitive research process.

When You Need More Certainty: AI Detection Tools

For situations where you're making high-stakes decisions (major hiring choices, large budget allocations based on competitor claims), you might want a second opinion from detection software.

Tools like Sensity, Optic (by Reality Defender), and even some built-in features in Google Photos can help flag potentially AI-generated imagery. They're not perfect, but they provide useful probability scores. Use these as verification after your manual checklist, not instead of it.

One important note: these tools work better on images that are obviously AI-generated. Recent Midjourney v7 images or DALL-E 3 outputs sometimes score lower on detection because the quality is higher. You still need human judgment.

The Bigger Picture: Why Your Team Needs This Skill

Teaching your team to spot fake images isn't about being paranoid about competitors. It's about building a culture of verification. Your marketing team should be checking competitor claims. Your hiring managers should be questioning too-perfect culture photos. Your sales team should verify customer testimonial images before referencing them in pitches.

This ties directly into the broader challenge of AI hallucination detection in your organization. Fake images are just one form of AI-generated misinformation. When you build verification habits for images, you're building verification habits across all content your team consumes.

Here's a practical step: in your next team meeting, show two images side by side (one real, one AI-generated) and have people guess. Most teams get surprised by how tricky this is. That's exactly why your team needs practice now, before high-stakes decisions depend on image authenticity.

If you're doing competitive research involving multiple image sources, consider using private search methods to organize and verify this information securely. Keep documentation of images you've verified and ones you've questioned. This builds institutional knowledge your team can reference.

FAQs

Can AI detection tools replace my own judgment?

No. These tools should augment your checklist, not replace it. They're most useful when you're already suspicious. If your manual review didn't catch anything, a detection tool might flag something you missed. But they also have false positives. Use them as a tiebreaker, not as your primary method.

What if my competitor's images are real but just look too good?

Good professional photography, excellent lighting, and professional editing can look unrealistic. The difference is: real photos will pass the hands, eyes, text, and physics checks. AI images fail at least one or two of these. If everything checks out, the image is probably real. Your competitor might just have hired a good photographer.

Should I accuse competitors of using fake images publicly?

Not unless you're 100% certain and it matters strategically. Instead, use the information internally. If competitor marketing claims are based on AI imagery, that tells you they might not have real case studies. That's valuable intel for your sales team. Document what you find, but verify thoroughly before any public accusation.

How do I train my team on this without creating paranoia?

Frame it as a verification skill, not a witch hunt. "Before we reference competitor metrics or try to match their marketing approach, we need to verify the source." Make it part of your standard due diligence process. This is especially important for hiring teams who might base decisions on company culture photos that don't represent reality.

The Real Competitive Advantage

Your competitors are getting better at creating fake content. But they're probably not training their teams to spot it from other companies. That gap is where you win. When you build verification into how your team consumes competitor information, you make better strategic decisions. You don't chase metrics that don't exist. You don't try to replicate case studies that were assembled in Figma. You compete on reality instead of impressions.

Next Wave Index can help you build these verification habits across your team's entire AI workflow, from image analysis to reporting to vendor evaluation.