I Called Your AI Agent and It Told Me Everything: Live Voice AI Red Teaming (<– add to your schedule)
You’ll leave this talk understanding:
– How voice AI agents are architecturally different from text chatbots
– The specific attack vectors unique to voice: transcription manipulation, DTMF injection, audio-layer prompt injection, and social engineering through vocal tone
– A repeatable methodology for testing voice AI systems in your own organization
– Concrete remediation strategies for the most common findings
Brian Cardinale:
Brian Cardinale is the Principal Security Researcher at SecureCoders and creator of the TEAPOT methodology for voice AI red teaming. He holds a CISSP and has spent his career breaking things that aren’t supposed to break. Brian built VoiceGoat, the first open-source vulnerable voice AI agent, and leads RedCaller’s research into adversarial testing of phone-based AI systems. He previously discovered CVE-2015-4670 and has conducted security assessments across financial services, healthcare, and critical infrastructure. When he’s not social engineering robots over the phone, he’s probably social engineering robots over the phone.