LLM으로 LLM 규정 준수 위험을 잡는 스타트업, 1,000만 달러 유치
Zerodrift는 AI가 생성한 지적된 메시지를 GDPR, SOC 2 등 표준에 맞게 LLM으로 재작성하는 규정 준수 레이어를 위해 1,000만 달러 시드 투자를 유치했다.
ai governance startup zerodrift secured $10m in seed funding from investors including a16z speedrun to launch a compliance layer that corrects model outputs before they reach users. the system uses deterministic code to identify violations of standards like soc 2 or gdpr and then invokes a separate llm to rewrite the message, ensuring compliance without the high latency of running a full guardrail model. ceo kumesh aroomoogan claims the funding round was oversubscribed by 3x and closed in three weeks, signaling strong enterprise demand for safety tooling. the technology initially targets consumer-facing chatbots but plans to expand into internal automated agent-to-agent communications.