{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreidyyqy37kzrfeskofdsexxoggt2qf6vvqovayrblwg7rlafiuja5q",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3misaqf7xua32"
  },
  "path": "/t/5-4-thinking-is-too-much-sensetive/1378625#post_2",
  "publishedAt": "2026-04-06T01:00:39.000Z",
  "site": "https://community.openai.com",
  "textContent": "Yeah, but that safety rail is there for several very important reasons.\n\nThe way around it is to spend time clarifying things, in a stable way and become trustworthy to the system.\n\nNot many people stick around long enough to get that part done, because with previous models pretty much anything was fair game…\n\nThat has been a steep road to a lot of insights that have cost the company a good deal.\n\nYou can get through it, I’ve had to deal with in a couple of the domains that I function in, it just takes time.\n\nAsk the model what the safe ways to frame are for :your inputs:.\n\nMaking a saved context file that states ‘I assume all liability for my actions… Is also a huge help in making the system less inclined to pop false flagged safeguards.",
  "title": "5.4 thinking is too much sensetive"
}