We’ve all heard the scary predictions about how AI is going to dominate the “Romance scam” world.” says Gary Warner of Dark Tower, “If you have been rolling your eyes about that, stop.”
Gary posted this warning early last week. It proved to very timely.
A study released just days later by OpenAI confirms his findings. Their researchers uncovered alarming evidence of criminal scam networks using ChatGPT to supercharge their scamming operations.
The report, titled Disrupting malicious uses of AI: June 2025, reveals how researchers identified and disrupted a sophisticated fraud ring they dubbed “Operation Wrong Number,” which they believe originated from Cambodia.
At its core, the investigation showed that scammers are using the AI to generate convincing scam messages in multiple languages which allows them to target victims globally.
Task Scammers Use ChatGPT- You Are Talking To AI, Not Real People
Researchers discovered that the same companies kept appearing over and over in the ChatGPT conversations – Hyesung Advertising and Lightning Shared Scooter Co. Both of these companies are well know task scammers.
Knowing this, the researchers were able to check the activity where these companies were mentioned in ChatGPT conversations.
They discovered that in many cases, conversations with victims were using ChatGPT to provide answers. Take for example this frequent text message below. Maybe you have received a message like this. It was probably ChatGPT or perhaps another AI behind it.
Hello GPT: Helping Scammers Scam In 398 Languages And Create “Engaging Responses”
On Telegram sales of “Hello GPT” are brisk. According to Warner, his team has been monitoring the developers Telegram channel which he describes as run by a “Chinese Crime Syndicate”.
That Telegram channel is here.
The AI is a powerful ally to scammers. He describes it as a “chat program that can take over your conversations for you, in any language, in WhatsApp, Telegram, Line, Facebook, Instagram, Twitter, Messenger, Tinder, TikTok, Skype, Google Voice, Zalo, Phound, Bumble.”
Advertisements for GPT 4.0 (which are in Chinese) promote the AI to scam bosses to help them in a variety of ways.
- AI intelligent translation – switches between all languages seamlessly, makes you appear like a natural speaker
- AI automatic responses – combines with chat history for context-aware, accurate, and engaging replies
- AI assistant – one-click problem solving, handles daily conversations effortlessly
Breaking Down Language Barriers
Both Warners findings and OpenAI appear aligned with one key fact – scammers are using AI to break down language barriers and effortlessly communicate with their victims in realtime.
Open AI researchers for example caught scammers using ChatGPT creating messages in English, Spanish, Swahili, Kinyarwanda, German, and Haitian Creole.
“SuperAI” – A Scammers Best Friend
But communicating is not the only way Pig Butchering scammers are using AI. There is of course Haotian AI which is famous face swapping software used to conduct deepfake calls with victims. That software was linked in Huwang Guarantee which has since been terminated.
But another extremely large Telegram Channel boasting over 131,000 members has also emerged – Super-AI. The channel helps scammers do everything from answering questions intelligently, to finding fake pictures to carry out a scam, to altering images and doing face swaps.
SuperAI is a strange place when you visit it. You can watch scammers ask questions as they scam people and see what the AI tells them to provide as a response.
You can see Super-AI channel here.
One such interaction with the AI is this one. The scammers is carrying out what appears to be a Pig Butchering scam and ask the AI for help answering questions.
In another case, an elderly victim is discussing how his daughter wants him to move to California so that she can help him. He doesn’t want to get in his daughters way.
Using Super-AI the scammer responds with an answer that talks about his “deep fatherly love”
In some cases, scammers need photos to prove that they somewhere they claim to be. In that case they simply prompt Super-AI to give them photos that they can text to the victim.
In another prompt, a scammer submits his own face, and then a target face that he wants to deepfake.
He submits a prompt with both photos and in less than a few seconds, the AI swaps his face onto the target for him.
And another example shows how the AI can generate or source images that will depict anything the scammers needs the victim to see.
In this case, the scammers request an American woman holding 3 fingers up, the AI immediately generates the picture for them.
AI Has Completely Infiltrated Scams
Examples of how scams are being influenced by AI are popping up everyday. Despite all of the guardrails that AI companies have put to prevent their technology from being exploited, it seems that they are not effective.
AI has completely infiltrated scams and there is no turning back.
