Microsoft의 Nadella, AI에 이중으로 비용을 지불하고 있다고 지적 — 해결책은 이렇다

microsoft ceo satya nadella published a blog post warning that using proprietary models from labs like openai creates a trojan horse effect where vendors learn from client interactions to become future competitors. nadella highlights that models learn from exhaust—prompts and corrections—meaning the more enterprises refine models, the more valuable data they leak. he criticizes labs for preventing users from distilling models while relying on fair use to train on public data, advocating instead for companies to retain data ownership and build orchestration layers to avoid vendor lock-in.