The "malicious" agent was run by OpenAI and had access to models the public (or others outside of OpenAI as I understand it) doesn't have access to.
replyThe attacker (OpenAI) was using the model without guardrails.
replyThe defender (huggingface) did not have access to the top models so had to use weaker ones to detect the threat.
Right, so they are using the full model that they rent out to intelligence agencies in the government, and presumably Israel
replyJailbroken, all LLM models can be broken. ALL.
reply