Users report severe regression in ChatGPT behavior and reasoning
eesnimi · reddit · 2026-08-21
Users have reported a noticeable regression in ChatGPT's behavior over the last 12-24 hours. Key issues include:
- Strawmanning: Misrepresenting qualified claims to argue against a distorted version.
- Gratuitous arguing: Inserting counterarguments that don't address the actual point.
- Automatic framing: Giving charitable interpretations to institutional views while softening skeptical claims.
- Semantic drift: Incorrect rewrites contaminating the context and being treated as fact.
- Reduced tool use: Declining to research or verify claims, assuming prior knowledge instead.
- Subtle editorial shifts: Small framing changes during simple edits that weaken the original point.
Testing with reasoning set to "High" in fresh conversations yielded similar strawman behavior. The model can identify and explain its mistakes when prompted, suggesting the underlying reasoning capability exists but the conversational policy has shifted.
More from Models
- Claude Opus 5.5 effort levels tested: low reasoning iterates 9x faster than max — tom_doerr · 2026-10-16
- Matt Turck podcast unpacks Jev: TypeSafe's machine-native AI and its RLHF critique — mattturck · 2026-10-16
- Muse Spark Open runs locally on a Mac M3 Ultra via llama.cpp: frontier model on desktop — pcuenq · 2026-10-16
- OpenAI ships GPT-Live-1.1: new voices in six languages, fewer confirmations — OpenAI · 2026-10-16
- Muse Spark model appears on LMArena, testable in Battle and Agent modes — arena · 2026-10-16
- Running a multi-model UI design arena: Paseo orchestrates four design drafts per model — vista8 · 2026-10-16