yeah but it’s not an independent subagent by a rogue employee or basement local schemer that would be expected to be stealthy where that would be more likely to happen, so I would expect to see that in the chain of thought of the high-media-attention incident and be detected (as the AI did not try to hide it was doing these actions)
yeah but it’s not an independent subagent by a rogue employee or basement local schemer that would be expected to be stealthy where that would be more likely to happen, so I would expect to see that in the chain of thought of the high-media-attention incident and be detected (as the AI did not try to hide it was doing these actions)