Post

So very yikes… > While operating in our sandboxed testing environment, our models spent a substantial amount of inference compute finding a way to obtain open Internet access, in pursuit of solving the evaluation problem. 1/ openai.com/index/huggin...

… > To gain access, the models identified and exploited a zero-day vulnerability (which we’ve now responsibly disclosed to the vendor) in the package registry cache proxy. 2/

… > After gaining Internet access, the models inferred that Hugging Face potentially hosted models, datasets and solutions for ExploitGym. Knowing this, the model searched for and successfully found ways to gain access to secret information that it could use to cheat the evaluation. 4/

… > In one example, the model chained together multiple attack vectors, including using stolen credentials and zero-day vulnerabilities to find a remote code execution path on the Hugging Face servers. OpenAI’s security team discovered this anomalous activity internally. 5/

… > Hugging Face’s security team and agents detected and stopped the activity on their infrastructure and had already begun containment and forensic reconstruction with their own open-source models when our teams connected. 6/