{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreidf5cx33fcuf72y6mzda5bm7b4nvgjrybq3cvrqraicbdl7fcdisi",
"uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3ml6rcxhyiou2"
},
"path": "/t/gpt-massive-anxiety-stress/1380366#post_4",
"publishedAt": "2026-05-06T12:47:25.000Z",
"site": "https://community.openai.com",
"textContent": "I disagree. Been coding w/ codex since inception. This behavior is exactly as OP described! I am beside myself with all the arbitrary editing, it now deletes unrelated code blocks, out of scope with instructions!\n\nin spite of prompts, and ignores REPEATED REQUESTS 10+ TIMES!?\n\nLOOKS LIKE WE’RE IN A TAILSPIN TO MEDIOCRITY…\n\nCONFIRMED\n\n * The model kept steering away from the user’s actual objective.\n * It substituted soft artifacts, outreach, summaries, and manual capture for executable proof.\n * It repeatedly failed to preserve the core requirement: automated, runner-recorded, sealed lower-boundary evidence.\n\n",
"title": "Gpt = massive anxiety+stress"
}