Something feels off. The photos are too perfect. They won't video call. You've been here before, or maybe this is the first time that gut feeling has kicked in. Here's how to find out if they are who they say they are.
Catfishing — using a fake identity online to deceive someone emotionally or financially — has become significantly more sophisticated since AI image and video generation became widely available. The tells that used to give catfishers away have been patched by AI tools that produce photorealistic faces, voices, and even video calls.
These are the most consistent warning signs that the person you're talking to may not be who they claim to be:
Their photos are too perfect. Studio-quality lighting on every photo, flawless skin, the same flattering angle in every shot. Real people's photo libraries are messy — candids, bad lighting, unflattering moments. AI-generated profile photos tend to look uniformly polished.
They avoid video calls. A person using AI-generated photos cannot video call using those photos. Excuses include poor internet, broken camera, shy personality, or simply repeated deflection every time you suggest it.
They have very few photos, all similar. Genuine people accumulate varied photos over years. A profile with only 4–6 professional-looking photos of the same person is a signal. AI generators produce distinct faces that can be hard to vary consistently.
The relationship escalated unusually quickly. Love bombing — intense affection, emotional dependency, and declarations of love very early — is a common precursor to financial requests in romance scams.
Their story has inconsistencies. A different city, a detail that doesn't match something they said before, a profession that shifts. Scammers managing multiple victims often lose track of the narrative they've given each person.
They've never been able to meet in person. Always a work emergency abroad, a family crisis, a deployment, a hospital stay. The meeting keeps getting delayed indefinitely.
They asked you for money or a favour. Any financial request — however reasonable it sounds — from someone you've never met in person is a strong signal of a romance scam.
If they've already asked for money
Stop all transfers immediately. Do not send more, even if they claim the first amount will be lost. Contact your bank to report the transactions. Romance scams cost victims worldwide over $1.3 billion annually according to the FTC.
The fastest check
How to check if their photos are AI-generated
If the person you're suspicious of has a public social media presence on X, Instagram, TikTok, Facebook, Reddit, or YouTube, you can verify their photos and videos in seconds without them knowing.
01
Find their public social media post
Go to their profile on X, Instagram, Facebook, or whichever platform you know them from. Find a post that contains one of their photos or a video of them. The post must be publicly visible — you should be able to see it without being logged in to the platform.
02
Copy the post URL
Click or tap on the specific post (not their profile page). Copy the URL from your browser's address bar. On mobile, tap the share icon on the post and select "Copy link".
03
Paste it into KweliAI
Go to kweliai.com, paste the link into the scan box, and press Scan. KweliAI extracts the image or video directly from the post and analyses it with three AI detection models simultaneously.
04
Read the verdict
You'll receive a clarity score between 0 and 100, a verdict (Real, AI-generated, or Deepfake), and a confidence rating. A high confidence AI-generated verdict means the image was created by an AI tool, not captured by a camera.
"KweliAI works silently from the post URL. You don't need to download anything, contact the person, or alert them in any way. The scan takes seconds and leaves no trace."
It's important to understand what a KweliAI result means — and what it doesn't.
An AI-generated verdict means the image or video was created by an AI tool such as Midjourney, DALL·E, Stable Diffusion, Sora, or Runway. The face in the photo does not belong to a real person — or if it is a real person, their appearance has been significantly AI-manipulated. This strongly suggests the profile is fake.
A real verdict means the image shows signs consistent with camera-captured media and no AI generation fingerprints were detected. This does not guarantee the profile is genuine — a real photo can still belong to a different person than they claim (identity theft). Combine this result with a reverse image search for a more complete picture.
A deepfake verdict means a real person's face or voice has been digitally manipulated using AI. This is different from a fully AI-generated face — it means they may have taken a real video and altered it.
Scan multiple posts from the same person for the most reliable picture. A pattern of AI-generated verdicts across several posts is a very strong indicator that the account is using synthetic content.
No public profile
What if they have no public social media?
If the person you're talking to claims to have no social media, or all their profiles are private, KweliAI cannot scan them directly — the tool requires a public post URL. However, you still have options.
Reverse image search. Save any photo they've sent you (screenshot if necessary) and upload it to Google Images or TinEye. If the photo appears elsewhere online under a different name, the identity is stolen.
Ask for a video call — specifically. Ask them to do something specific on camera in real time, like hold up a piece of paper with your name written on it, or wave with their left hand. A pre-recorded or AI-generated response cannot respond to a live, specific request.
Ask them to post something publicly. "Can you post a selfie on your X account so I can see your real profile?" A genuine person will usually be happy to. Someone using a fake identity will find reasons not to.
Search their claimed name + photos. Search their full name on Google with "photos" or "images". If you can find a public profile, you can then use KweliAI on one of those posts.
Next steps
What to do if KweliAI says they are fake
Getting a confirmed AI-generated or deepfake result is difficult to process emotionally, especially if you've invested time and feelings in the relationship. These are the practical steps to take.
Stop all contact and money transfers immediately. Do not send more money, gift cards, cryptocurrency, or wire transfers. Do not respond to requests claiming the first transfer is at risk.
Do not send any more personal photos or videos. These can be used for sextortion — blackmail using intimate images — even after you've cut off contact.
Report the profile to the platform. Use the Report button on the profile. This helps protect others from the same scammer.
Report to your country's fraud authority. In the US, report to the FTC at reportfraud.ftc.gov. In the UK, contact Action Fraud at actionfraud.police.uk. In Kenya, contact the Directorate of Criminal Investigations (DCI).
Contact your bank if money was transferred. Banks have fraud departments that may be able to reverse or trace recent transfers. Act as quickly as possible — delays reduce the likelihood of recovery.
Talk to someone you trust. Catfishing is designed to isolate victims. Reach out to a friend, family member, or support organisation. You are not the first person this has happened to and it is not your fault.
Common questions
Questions people ask when they think they're being catfished
Can I check if someone's photos are real without them knowing?
Yes. If they have any public posts on X, Instagram, Facebook, TikTok, Reddit, or YouTube, paste the post link into KweliAI at kweliai.com. The scan is completely silent — they receive no notification. KweliAI works from the public post URL without requiring any interaction with the account.
What if the photos come back as real but I'm still suspicious?
A real verdict means no AI generation signals were detected in that specific image. It doesn't prove the person is who they claim to be — real photos can still belong to a different person (identity theft from someone else's social media). Combine a real verdict with a reverse image search to check if the photo appears under different names elsewhere on the internet.
My match says they don't have social media. Should I trust them?
Having no social media presence is unusual but not impossible. Combined with other warning signs — refusing video calls, asking for money, inconsistent stories — it significantly increases the likelihood of a fake identity. Ask them to create a temporary public post so you can verify they are real.
They video called me once. Does that mean they're real?
Not necessarily. Deepfake technology can now generate real-time video of AI faces on a call. However, real-time deepfake calls typically have quality issues, slight delays, and struggle with unexpected requests (like turning sideways or holding something up to the camera). If the call felt slightly off or they reacted strangely to unexpected movements, it may still be synthetic.
How many posts should I scan?
Scan at least 3–5 different posts if possible. A single AI-generated verdict is significant. Multiple AI-generated verdicts across different posts from the same person is very strong evidence the account is using synthetic content. Free users have 5 scans per day — enough to verify a profile thoroughly.