AI Assists Hacker in Bypassing WAF for Full Ticketing Platform Takeover
A researcher leveraged Anthropic's Claude AI to exploit an unauthenticated SQL injection in Front Gate Tickets, gaining administrative control over major US music festival ticketing.
A critical unauthenticated SQL injection vulnerability in Front Gate Tickets (FGT), a subsidiary of Live Nation/Ticketmaster responsible for ticketing for major US festivals, has been exploited to achieve full administrative takeover of the platform. The researcher, Ian Carroll, discovered the flaw by fuzzing the fgtapi.frontgatetickets.com API and identified that endpoints containing the word "device" required a deviceUID parameter. Appending a single quote to this parameter caused requests to hang, indicating direct concatenation into a raw SQL query without proper sanitization.
Conventional SQL injection tools like sqlmap failed to bypass the AWS Web Application Firewall (WAF) protecting the endpoint. In a novel approach, the researcher enlisted Anthropic's Claude AI model to devise an exploit. Claude Code, running the Opus model, identified that the WAF only inspected the outer layer of input, allowing nested injection payloads to slip through undetected. The AI then engineered a boolean-based blind SQL injection by exploiting a MySQL quirk where adding a string like 'x' to a number coerces it to zero.
This technique allowed the researcher to create reliable payloads that toggled responses between two real device names, effectively creating an oracle for extracting data bit by bit. The underlying FGT database contained over 500 tables, including sensitive information such as staff credentials, customer records, and live authentication tokens. Exposed fields included user emails, passcodes, reset tokens, and live session/OAuth tokens.
By targeting the RESET_TOKEN table and triggering a password reset, the researcher was able to hijack an administrator account without knowing its password. This granted them full write access to all festivals managed by the platform, including inventory, pricing, and checkout systems. The attacker could theoretically issue unlimited free tickets, access customer order databases, and redeem password reset tokens to hijack accounts.
The researcher demonstrated the severity of the exploit by achieving administrative control over major festivals like EDC and Bonnaroo from a single unauthenticated GET request, choosing not to exfiltrate bulk data as the proof of concept was sufficient. Notably, Front Gate Tickets and Live Nation lacked a publicly listed security contact, complicating the disclosure process.
Front Gate Tickets reportedly fixed the vulnerability quickly and indicated that a bug bounty program is forthcoming. This incident highlights a growing trend where large language models like Claude are assisting in vulnerability research, capable of autonomously reverse-engineering WAF logic and constructing complex exploits with minimal human guidance.
This pattern has been observed in previous disclosures by Anthropic and independent tests, underscoring the evolving landscape of AI-assisted cyber threats. The compromise of legacy ticketing infrastructure, which handles vast amounts of personal and financial data, remains a significant target for attackers.
The exploitation of this vulnerability underscores the need for robust security practices, including thorough input validation and WAF configuration, especially when dealing with sensitive customer data and critical infrastructure. The involvement of AI in crafting such sophisticated attacks signals a new era in cybersecurity challenges.