Researcher Uses Claude AI to Unlock Backdoor Affecting Tickets for Most US Music Festivals

Security researcher Ian Carroll made an unsettling discovery in April: using Anthropic's Claude Opus 4.7, he found a way to gain complete administrator access to Front Gate Tickets—the company that manages ticketing for nearly every major US music festival, including Lollapalooza, South by Southwest, and Austin City Limits. Once inside, he could have freely generated premium VIP passes for himself and others to any event, with no restrictions.

Carroll, who runs the startup Seats.aero and conducts independent security work, never actually issued bogus tickets. Instead, he reported the vulnerability to Front Gate, which has since patched it. The incident underscores how capable modern AI tools have become at unearthing exploitable weaknesses across digital infrastructure.

How the Vulnerability Worked#

Carroll began investigating Front Gate after noticing the company controlled ticketing for practically every major US festival except Coachella—making it essentially the Ticketmaster of festival ticketing. When he probed the company's website, he discovered what appeared to be a SQL injection vulnerability, a common flaw where attackers input malicious commands into text fields to manipulate backend databases.

The site's web application firewall initially blocked his attempts to exploit it. He then turned to Claude and asked it to find a bypass. The AI quickly delivered: it identified that a "nested SQL query"—a query embedded within another query—could slip past the firewall's detection. Claude then generated a script that exposed data from 500 databases containing customer information (names, emails, mailing addresses) and staff records.

Armed with staff access, Carroll located a super-administrator account, used the password reset mechanism to obtain the reset code stored in the backend, and took over the account with a single new password—no two-factor authentication required. From there, he could add any ticket he wanted to a shopping cart as complimentary passes, with no system safeguards preventing issuance.

What the AI Could Do on Its Own#

Carroll expressed surprise at Claude's autonomy in solving the exploit. "I think there's a very good chance it could have found this exploit end-to-end without me doing anything at all," he reflected. He noted that for the first time, he possessed a vulnerability he didn't fully understand because Claude had written the bypass technique itself.

Industry and Company Response#