Estimating worst case frontier risks of open weight LLMs

Researchers studied worst-case risks of releasing open-weight large language models by conducting malicious fine-tuning (MFT) experiments on gpt-oss. The study specifically examined how fine-tuning could maximize dangerous capabilities in biology and cybersecurity domains.

• →Researchers introduced malicious fine-tuning (MFT) as a method to assess maximum risk potential of open-weight LLMs.
• →The study focused on two high-risk domains: biology and cybersecurity capabilities.
• →Open-weight model releases face scrutiny over potential misuse through targeted fine-tuning.
• →The research aims to quantify frontier risks before public model releases.
• →Fine-tuning techniques can potentially unlock dangerous capabilities in publicly available models.