OpenAI says it has shut down a coordinated scam campaign operating out of Asia that was hijacking multiple ChatGPT accounts to lure victims into fraud schemes. The attackers weren’t trying to jailbreak the model for fun — they were using ChatGPT as a scalable social‑engineering engine, generating convincing outreach messages, fake business pitches, and personalized hooks designed to reel in targets across messaging apps and email.
Image Courtesy : cnbc.com
The campaign relied on a network of compromised or fraudulently created ChatGPT accounts. By distributing activity across many logins, the scammers tried to avoid detection while pumping out large volumes of polished, human‑like messages. OpenAI’s security team eventually traced the pattern, identifying clusters of accounts behaving in sync and producing content tied to known scam templates.
Once flagged, the accounts were banned, cutting off the attackers’ access to ChatGPT’s generative tools. OpenAI says the takedown is part of a broader effort to stop malicious actors from weaponizing AI for phishing, impersonation, and financial fraud. The company has been tightening automated detection systems and monitoring for unusual usage spikes — especially those that mimic coordinated campaigns.
The incident highlights a growing reality: scammers are increasingly turning to AI not just to write messages, but to scale operations, adapt scripts to different languages, and tailor outreach to specific victims. Cutting off access is only part of the solution; platforms now have to anticipate how attackers will evolve.
For everyday users, the takeaway is simple but sobering — the next scam message you receive might not come from a human at all. And stopping these campaigns requires constant vigilance from AI companies, not just reactive bans.