OpenAI Shuts Down Cambodia-Based ChatGPT Scam Network Behind Romance, Investment and Impersonation Fraud

OpenAI Dismantles Cambodia-Linked ChatGPT Scam Network Behind Romance, Investment and Impersonation Fraud

Organised crime groups are no longer just stealing identities or sending phishing messages.

They are increasingly turning artificial intelligence into a tool that helps run large-scale fraud.

OpenAI has now shut down a coordinated network of ChatGPT accounts linked to a criminal operation in Cambodia that used the technology to support romance scams, fake investment schemes, gambling fraud and law enforcement impersonation.

The investigation began after WhatsApp shared intelligence that led OpenAI to uncover a network believed to have operated from Cambodia's Preah Sihanouk province, an area repeatedly linked by public reports to online scam compounds and human trafficking.

According to OpenAI, the accounts showed clear signs of coordinated activity designed to deceive victims across multiple types of fraud.

How ChatGPT Became Part of a Criminal Operation

Rather than using ChatGPT for isolated scams, the group integrated the AI into nearly every stage of its operation.

OpenAI said the accounts relied on its models to generate and translate conversations, create convincing fake online identities, produce promotional content for fraudulent schemes, and assist with routine administrative work inside the organisation.

The company said the operation "may have interacted with hundreds of targets across multiple scam types", although the total financial damage remains unknown.

OpenAI said,

"The full scale of financial losses associated with the network is unknown, but based on the scammers' own communications, the operation may have interacted with hundreds of targets across multiple scam types. User conversations referenced individual victims losing thousands of dollars, although we are unable to independently verify those claims.”

Four Scams Running Side by Side

The criminal network did not rely on a single fraud model.

Instead, it operated investment scams, romance fraud, fake gambling promotions and impersonation scams at the same time, often combining them to increase the chances of success.

Victims were first approached through messaging platforms such as WhatsApp and Telegram before conversations gradually shifted towards financial opportunities or fabricated legal problems.

Romance scams encouraged emotional attachment before directing victims towards fake cryptocurrency or precious metals investment platforms.

Other accounts posed as gambling platform customer support, luring users with fake bonuses or prize notifications.

A scammer from the crime ring used ChatGPT to build a fake cryptocurrency trading interface. (Source: OpenAI)

Another part of the operation impersonated law enforcement officers, falsely accusing victims of criminal offences and demanding payments to settle fabricated fines.

Inside The Ping Zing And Sting Strategy

OpenAI described the network's approach as a three-stage process known internally as "Ping", "Zing" and "Sting".

The "Ping" stage focused on making first contact, with ChatGPT helping scammers translate messages and generate natural conversations in multiple languages.

The "Zing" stage was designed to build trust through emotional manipulation.

Victims were promised guaranteed profits, "risk-free" investments and exclusive opportunities while being encouraged to keep conversations private and act before fake promotions supposedly expired.

The final "Sting" stage involved extracting money.

Victims were instructed to transfer funds to unlock fabricated rewards, pay activation charges or settle non-existent fines.

They were then asked to provide screenshots of completed payments or share account information as proof.

Fake Documents Added Another Layer Of Deception

OpenAI also discovered that the network used ChatGPT to help produce fraudulent documents that made scams appear more legitimate.

These included fake passports, legal notices, stock purchase confirmations and screenshots showing counterfeit gambling platforms.

The fabricated materials were designed to convince victims they were dealing with genuine organisations or official authorities.

Beyond interacting with victims, the AI was also used to maintain fake social media profiles, research dating platforms, translate communications and produce marketing material that supported the wider scam operation.

Links To Human Trafficking Raise Further Concerns

The investigation also uncovered conversations pointing to possible links with organised crime groups involved in human trafficking and forced labour.

Some accounts generated recruitment advertisements promising flights, accommodation, meals, visas and work permits to attract workers to Cambodia.

Scammers in the network used AI-generated images to advertise Cambodia job offers on social media. Redactions added by OpenAI. (Source: OpenAI)

ChatGPT was later used to translate internal communications and draft notices covering recruitment, immigration status, salary deductions, disciplinary fines, employee debts and work conditions.

Some conversations referred to detention, failed escape attempts and fears of criminal prosecution.

While OpenAI said these records do not prove what happened to specific individuals, they closely resemble patterns documented in public reporting on Southeast Asian scam compounds, where trafficked workers are forced to participate in online fraud.

Account Bans Alone Cannot End AI-Driven Fraud

OpenAI has banned the accounts involved and shared threat indicators with law enforcement agencies and industry partners to help prevent similar activity.

However, the company acknowledged that removing accounts addresses only one part of a much larger problem.

Organised criminal groups can migrate to different AI platforms, create new accounts and modify their methods while continuing to operate.

As generative AI becomes more capable, criminal organisations are also finding new ways to automate deception, allowing smaller teams to run larger fraud operations with greater efficiency than before.

The Fight Against AI Abuse Is Becoming More Complex

Coinlive believes this case highlights a growing challenge facing the AI industry.

Tools like ChatGPT are designed to improve productivity and communication, yet the same capabilities can also be exploited to scale fraud when they fall into the wrong hands.

Preventing abuse will require more than account bans.

It will depend on continued cooperation between AI companies, technology platforms and law enforcement to stay ahead of organised criminal networks that adapt as quickly as the technology itself.