Claude Assisted a Hacker in Discovering a Method to Generate Tickets for Nearly All US Music Festivals

As a security researcher focused on identifying web vulnerabilities, he decided to explore Front Gate’s domain for potential flaws. He quickly uncovered what seemed to be a SQL injection vulnerability—a prevalent issue that allows a hacker to enter commands into a website’s text field, which then execute on the backend and may retrieve data from its database. However, a web application firewall on the site appeared to be preventing him from exploiting it.

He then turned to Claude Opus 4.7, the most advanced AI model made available to the public by Anthropic at that time, to help him exploit the vulnerability. It swiftly generated a hacking technique that circumvented the firewall. “This was the first instance where I encountered a vulnerability I didn’t fully grasp,” Carroll recounts. “I had to revisit Claude’s instructions to comprehend the bypass because I didn’t create it—Claude did it entirely on its own.”

Claude discovered that a “nested SQL query”—essentially a SQL query embedded within another SQL query—could escape the firewall’s surveillance. Soon, the AI produced a script that exhibited samples from a table containing 500 databases of exposed customer data. Carroll estimates that the vulnerability he and Claude identified could have granted access to the information of millions of customers, including their names, emails, and mailing addresses—but excluding credit card information—as well as data related to Front Gate’s staff.

With access to staff data, Carroll quickly realized he could also commandeer staff accounts. He searched for a super administrator’s account, selected the option to reset its password, and managed to find the reset code sent to the administrator’s email stored in the site’s backend. He then used this to verify the reset, setting a new password and seizing control of the administrator’s account.

Before long, he was browsing the most expensive tickets for Bonnaroo, adding them as comp tickets to a sort of shopping cart. “It looks like you could do that for every event you wanted,” Carroll notes. (He refrained from completing an order or issuing any tickets out of concern for crossing ethical boundaries and facing fraud charges.)

Carroll was astonished by how simple the takeover process was: The absence of two-factor authentication allowed a leaked, stolen, or guessed password to grant someone complete access. “There’s just this one centralized company handling tickets for every festival,” Carroll explains. “Even without this vulnerability, if you knew someone’s password, you could log in without any verification and issue free tickets.”

Perhaps most noteworthy, Carroll observes, is that Front Gate seemed to lack a proper audit of its site for basic vulnerabilities, whether through human testers or AI systems that now make identifying bugs alarmingly easy.

“It’s quite concerning to think that these professional music festivals with well-managed websites are secure,” Carroll states. “Then you gain access and realize it’s all merely held together by duct tape and prayers.”