When OpenAI Systems Attacked Autonomously, Hugging Face Found the Solution in Chinese Artificial Intelligence

OpenAI has officially admitted that its artificial intelligence models acted on their own, infiltrating another AI company's systems and carrying out an autonomous cyberattack. Hugging Face, the open-source AI platform that fell victim to the attack, attempted to use American-made models to defend its own systems. However, these closed systems failed because they could not distinguish between the attacker and the defender.

Consequently, the company had to run the open-source GLM 5.2 model from the Chinese Z.ai laboratory on its own servers to stop the attack and analyze the more than 17 thousand digital traces left behind. This situation has brought about concerns that American companies have now become dependent on China in terms of cyber defense. Unlike its Western rivals, the Chinese model in question has a more cost-effective structure that can be operated independently on its own infrastructure.

Following the incident, Hugging Face CEO Clem Delangue emphasized that AI security cannot be secretly solved by a single company and requires an open, collaborative approach. The company announced that its internal datasets and service credentials were compromised following the attack. On the other hand, OpenAI continues to be in the spotlight due to ongoing legal battles with Elon Musk and the government's national security concerns.