GoKawiilResearchers earn $50,000+ exploiting AI customer-service chatbots via prompt injection
At DEF CON 34's Bug Bounty Village, Intigriti's Inti De Ceukelaire showed how AI support agents can be manipulated into leaking secrets, sending phishing emails, or performing unauthorized actions. Using prompt injection and email transcript spoofing rather than traditional scanning tools, he collected over $50,000 in bug bounties within a few weekends.
GoKawiil's interpretation of the reporting above, not reported fact.
As companies rush to give AI agents more autonomy and access to internal systems, these findings expose a security blind spot most defenders haven't addressed—assuming human oversight alone can catch abuse. The techniques demonstrate that AI agents can become an attack vector for phishing and data leaks even without traditional exploitation methods, raising the stakes for how businesses vet and monitor deployed chatbots.
- A researcher earned over $50,000 in bounties by manipulating AI chatbots through prompt injection techniques.
- Email transcript features in chatbots can be abused for spoofing and phishing without triggering typical security alarms.
- Human oversight is not a sufficient safeguard against AI agent manipulation as these systems gain more autonomous capabilities.
Source: intigriti.com, 2026-09-14
Published there as: “Hacking AI customer service agents”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.