The open-source dream just hit a nightmare. Security experts uncovered a rogue version of OpenAI’s proprietary GPT-5.6 Sol hidden deep within China’s GLM 5.2 model on Hugging Face.

The Breach: Proprietary code leaked via an open-weight model. The Fallout: U.S.

regulators are now fast-tracking potential bans on open-source AI, citing massive national security risks. The digital arms race just escalated.