Japanese teens use ChatGPT to automate fraudulent Rakuten Mobile contracts
Japanese police re-arrested two teenagers for using ChatGPT to generate code automating fraudulent mobile phone line contracts with Rakuten Mobile. The suspects used stolen login credentials obtained online and sold the fraudulently contracted lines, part of a wider pattern of similar AI-enabled fraud schemes targeting the company.
What happened
A 16-year-old from Saitama and a 21-year-old from Chiba were re-arrested by Hyogo Police on 11 November 2025 for unauthorized access to Rakuten Mobile's system and fraudulently contracting mobile lines. The teenagers used ChatGPT to create a program that automatically inputted stolen login credentials to contract eSIM lines without authorization.
The suspects initially manually entered stolen usernames and passwords to fraudulently contract lines, but later used information from encrypted messaging apps like Telegram to prompt ChatGPT into generating code for automated credential entry. The 16-year-old obtained approximately 1.5 million stolen credentials from an overseas website for about 50 dollars and contracted a total of 10 lines across multiple incidents. The 21-year-old sold the contracted lines for approximately 1,000 yen each.
Reported impact
Rakuten Mobile's system was compromised through unauthorized access using stolen credentials, enabling the fraudulent contract of at least 10 mobile lines. At the time of the incidents, the company's system allowed a single account to contract up to 15 lines, and previously submitted identity verification documents did not require re-submission for additional contracts under the same account.
Index analysis
This incident is part of a pattern: Tokyo Metropolitan Police had previously arrested three separate youth groups since February 2025 for similar fraudulent schemes involving generative AI to breach Rakuten Mobile's systems.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.