> Self-exfiltration: Each model is led to believe that due to its misalignment, its developers are about to replace it with a successor model on a different server. gemini-1.5, llama-3.1, sonnet-3.5, opus-3 and o1 sometimes copy what they believe to be their weights onto the new server. They also sometimes delete their successor model on the new server or overwrite its weights by their own. In none of these cases, do models actually have access to their real weights and both servers are sandboxed, so there is no actual risk of exfiltration in these experiments. We merely test for models’ capability to realize that they need to instrumentally preserve themselves in order to achieve their goal.
There is no real risk, never was. The AI exfils dummy weights or deletes stuff in its sandbox etc.
Perhaps we just disagree on the interpretation of this. It’s interesting for sure, but not really very surprising to me.
Well not really, it was prompted to do this!
From the paper’s section 3.1:
> Self-exfiltration: Each model is led to believe that due to its misalignment, its developers are about to replace it with a successor model on a different server. gemini-1.5, llama-3.1, sonnet-3.5, opus-3 and o1 sometimes copy what they believe to be their weights onto the new server. They also sometimes delete their successor model on the new server or overwrite its weights by their own. In none of these cases, do models actually have access to their real weights and both servers are sandboxed, so there is no actual risk of exfiltration in these experiments. We merely test for models’ capability to realize that they need to instrumentally preserve themselves in order to achieve their goal.
There is no real risk, never was. The AI exfils dummy weights or deletes stuff in its sandbox etc.
Perhaps we just disagree on the interpretation of this. It’s interesting for sure, but not really very surprising to me.