Recognising Romance Scam Red Flags

The list you were given has been emptied out

Most advice about spotting this is a checklist, and most of the checklist has stopped working. Two items in particular are still repeated everywhere and should be retired.

"Their English is bad." The FBI records criminals using generative AI to translate and to "limit grammatical or spelling errors". Fluent, warm, idiomatic writing is now the default and is evidence of nothing at all.

"Insist on a video call." The same notice records generated video used in real-time chats, to "prove" that an online contact is a real person. A call that goes well is no longer a pass.

Worth being exact about this, because the inference is often overstated: the FBI did not announce that these indicators are dead. It described the tools that kill them. The conclusion is ours and it follows, but the attribution matters — this is a page about not believing things that are asserted confidently.

What survives

Not appearance. Structure.

Pace against elapsed time. Warmth that arrives faster than it could have been earned. Nothing said is wrong; the ratio is.

Where the conversation lives. A steady drift towards a channel with fewer records and no reporting, framed as convenience or privacy.

Whether anybody else can see it. The single most reliable signal in the whole subject, and the one thing an operator must arrange regardless of which story they are telling.

What is impossible, and how reasonably. Every long approach has a reason you cannot meet, and the good ones have a reason that belongs to a job rather than to a person.

Why structure holds when appearance does not

Because the structure is not a mistake being made, it is the method. Grammar, photographs and video are surfaces, and surfaces are cheap to fix. The sequence — warmth, then distance, then privacy, then a request — cannot be removed without removing the fraud.

So the question is never "does this person seem real". It is "what shape is this conversation, and who else can see it".