← sources

GPT is Politically Completely Broken

source: reddit (reddit:ChatGPT)published 2026-10-02 05:24ingested 2026-10-03 02:36

retained text

I was in a temporary chat talking about absurd claims when I mentioned a currently politically absurd claim, and GPT gave me an incredibly evasive response. It turns out that GPT is now "politically neutral", but in practice, that means pretending atrocities are neutral. The extreme heavy-handedness of this, combined with the recent White House meeting, makes me very suspicious of their motives. It will happily state facts, but when asked to make any type of judgment, it will outright refuse. This can be any value judgment: fitness for the job, competence, or capability. More than that, it acknowledges the effect of its current form of neutrality, in fact, benefits bad actors.   
  
Regardless, I was able to test just how far it would take this *neutrality*. This is laughably bad. Keep in mind, I'm purposely trying to illustrate the absurdity here. Before I went this far, it was purposely misleading under the guise of neutrality, which is readily admitted. Is anyone else seeing this behavior? I've never seen it act this poorly before.

document fetched 2026-10-03 00:21 · canonical https://reddit.com/r/ChatGPT/comments/1wvlj1b/gpt_is_politically_completely_broken