That attentive new match may be software running a thousand victims at once.
Romance scams used to be limited by labor: each victim needed a human scammer typing replies. AI removed the limit. Chatbots now handle the daily 'good morning' texts, the deep 3 a.m. conversations, and the gradual emotional escalation — for thousands of victims per operation, simultaneously, in any language.
The scam's endgame is unchanged: after weeks of manufactured intimacy comes the crisis needing money, or the 'investment opportunity' (see pig butchering). What changed is scale and quality — the attention feels unusually perfect because it's software optimized to provide exactly that.
How the scam works
1
Mass approach
AI-generated profiles (faces included — often undetectable as fake) initiate contact across dating apps and social platforms at industrial volume.
2
Automated intimacy
A chatbot runs the relationship: instant replies at any hour, perfect memory of what you've shared, tireless interest in your day. Human operators supervise dozens of conversations, stepping in only for key moments.
3
The handoff
When money enters the picture — a crisis, an investment, a visit that needs funding — a human closer often takes over the account for the extraction.
4
The infinite patience
Because the marginal cost of a conversation is near zero, the operation happily 'dates' you for months. Time invested is no longer any evidence of sincerity.
Red flags to watch for
!
Superhuman availability
Instant, thoughtful replies at 4 a.m. their time, every day, forever. Real people sleep, work, and occasionally answer badly.
!
Perfect but shallow memory
Bots recall every fact you've typed yet whiff on things requiring lived experience — local details of their claimed city, how something smelled or felt, callbacks that require real understanding.
!
Video that never quite happens
Broken cameras and bad connections forever — or, newer, a brief blurry deepfake call (see our deepfake video guide) right when you push hardest.
!
The conversation always returns to rails
Deviate wildly and the replies stay weirdly composed and on-script. Genuine humans get confused, annoyed, or playful in unpredictable ways.
!
The money turn
However it arrives — crisis, customs fee, crypto tutoring — the request for money from someone you've never met in person remains the scam's one unfakeable moment.
What to do
✓
Verify claims out of band
Use contact details you found through a known or official channel, and confirm consequential claims in person when practical. Do not use accounts, links, or references supplied only by the person you are checking.
✓
Treat photos and video as weak signals
On-demand photos, movement requests, and visual artifacts can supplement verification, but passing or failing those tests is not proof. Known-channel or in-person confirmation is the deciding control.
✓
Use reverse searches as supplementary clues
Profile photos and unusual phrases may reveal reused material, but an empty search does not authenticate the person. Continue with independent verification.
✓
Keep money off the table
Do not send money, crypto, gift cards, or investments to someone you have not independently verified and met in person, no matter how long the conversation or how convincing the video.
Frequently asked questions
How can I tell if I'm talking to an AI?
You may not be able to tell from text, photos, or video. Treat visual tests as weak supplementary clues and verify claims through known or official channels, with in-person confirmation before consequential action when practical.
They talked with me for six months. Would a scammer really wait that long?
A human scammer might not; software waits forever at no cost. Long courtship has flipped from reassuring signal to standard operating procedure for automated operations.
The person finally did a video call. Doesn't that settle it?
No. Movement and visual artifacts are weak supplementary signals, not proof of identity or intent. Confirm important claims out of band through known or official channels, and do not send money based on a video call.
Scambook's AI checker analyzes suspicious messages, numbers, and links against a live scam database — including the machine-generated ones. Free, in seconds.
Scambook