Like, if someone asked it to explain differing violent crime rates in America based on race and one of the pathways the CoT takes is that black people are more murderous than white people. Even if the specific reasoning is abandoned later, it would still be ugly.
Although maybe AIs will end up with a more sophisticated take on the problems than your average human.
For something usually hidden the first two don't really apply that well, and the last would have to be really blatant unless you want an article about "Model recovers from mistake" which is just not interesting.
And in that scenario, it would have to mean the CoT contains something like blatant racism or just a general hatred of the human race. And if it turns out that the model is essentially 'evil' but clever enough to keep that hidden then I think we ought to know.
https://techcrunch.com/2024/09/12/hacker-tricks-chatgpt-into...