Microsoft CEO Satya Nadella warns of AI-driven IP theft, calls for new patent model

Satya Nadella, Microsoft’s chief executive, used an X post to describe what he calls the “reverse information paradox,” where enterprises that pay for AI services also relinquish proprietary knowledge in the prompts and corrections they provide. He warned that this data leakage could allow AI providers to “steal” intellectual property, urging companies to create a hard trust boundary that keeps learning loops inside their own tenant.

Nadella echoed concerns raised by Palantir CEO Alex Karp about AI models becoming Trojan horses for corporate IP. He cited examples such as Anthropic’s expansion into new product categories after training on partner data. To address the risk, Nadella proposed a new AI‑specific patent system that would protect the ideas disclosed to AI models, ensuring that firms retain ownership of the knowledge they contribute.

The call for stronger data‑ownership mechanisms aims to preserve enterprise demand for large‑language‑model services while safeguarding core business secrets.