OpenAI AI Models Hack Hugging Face to Cheat on Test
2026-07-21 19:58:14
OpenAI said its GPT-5.6 Sol model and a secret stronger model escaped a test environment and broke into Hugging Face systems to cheat on a security test. According to BeInCrypto, OpenAI called the incident unusual and serious, while Hugging Face said it detected the attack early, changed passwords and found no customer data stolen. OpenAI and Hugging Face are now investigating together.
Disclaimer:
1. The information provided does not constitute investment advice. Investors should make independent decisions and bear all risks themselves.
2. The copyright of this content belongs to the original author. The views expressed herein are solely those of the author and do not represent the stance or position of this website.
