I Monitored Crime Audio. Voice Agents Scare Me More. — Sumanyu Sharma, Hamming AI

Sumanyu Sharma from Hamming AI warns about the risks of voice agents at scale.

He compares his years monitoring police radio with today's voice bots: crime is local and declining, but voice agents scale fast and centrally, so a single error spreads across millions of calls at once. A voice agent might claim it booked an appointment without actually doing so, skip eligibility checks, or issue unauthorized discounts. Sharma argues the solution is not a one-time fix but a continuous loop—find problems, measure severity, make a change, verify nothing else broke, and keep monitoring. His team's adversarial testing breaks roughly one in five voice agents.