Meta deletes 6.8 million fraudulent WhatsApp accounts, thanks in part to OpenAI.

In the first half of 2025, Meta announced it had deactivated more than 6.8 million WhatsApp accounts identified as fraudulent, a large-scale operation aimed at protecting the platform's more than two billion users from scam attempts orchestrated by international criminal networks. According to a post on the company blog , most of these accounts were traced back to scam centers based in Southeast Asia, where organized groups use social engineering and automation techniques to approach victims through seemingly innocuous messages.

The scams

Meta's analysis indicates that fraudsters created profiles with low-cost virtual numbers, often masquerading as trusted contacts such as friends, colleagues, or public service workers, to then trigger requests for payments, sending verification codes, or installing malicious apps. By monitoring behavior patterns and cross-referencing them with databases of abuse signals, WhatsApp's artificial intelligence systems quickly identified anomalies such as abnormal message sending rates, the simultaneous use of hundreds of profiles from the same IP address, and repeated attempts to add users to spam-heavy groups.

Collaboration with OpenAI

Meta emphasizes that these measures not only reduce the number of overt scams, but also discourage criminals from using WhatsApp as a vehicle to extort money, steal personal data, or spread malware. The company collaborates with international law enforcement agencies and private intelligence platforms to map and dismantle networks based in countries such as Indonesia, Malaysia, and the Philippines, where the sale of virtual SIM cards and cheap digital labor fuel a thriving online fraud market.

Most importantly, Meta confirmed it collaborated with OpenAI to expose a center in Cambodia where various scams were organized, including payments for fake likes, scooter rental pyramid schemes, and advertising for cryptocurrency investments. The scammers used ChatGPT to generate the initial text message containing a link to a WhatsApp chat, then redirected the victim to Telegram, where they were tasked with liking TikTok videos. The attackers attempted to build trust by explaining how much the victim had supposedly already "earned," before asking them to deposit money into a cryptocurrency account as the next step in the scam.

The new tools

Meta has introduced a series of tools and features to strengthen user defenses. These include a "suspicious login" alert, which will immediately notify users if someone attempts to register from an unused device or number; an updated spam filter, capable of blocking messages containing links known to be phishing; and the ability to more easily monitor active sessions from connected devices, allowing any unauthorized logins to be disconnected with a single tap.

The challenge to scammers remains, however. For this reason, the Menlo Park company urges users to keep the app updated, enable two-step verification to protect their accounts, and be wary of unsolicited messages requesting urgent action.