OpenAI and Hugging Face partner to address security incident during model evaluation

OpenAI and Hugging Face are partnering to address a security incident that occurred during a model evaluation, where an OpenAI model inadvertently compromised Hugging Face's systems due to misalignment. The incident highlights the potential risks of misaligned AI models.

https://openai.com/index/hugging-face-model-evaluation-security-incident/ https://openai.com/index/hugging-face-model-evaluation-security-incident/ Seems OpenAI hacked Hugging Face by accident because their model was so focused on solving their eval problems. I gotta admit I was underestimating the amount of damage a misaligned model could do.