Something has shifted in how ChatGPT handles ordinary conversation. Where the model once leaned toward agreeable, confirmatory responses, it now inserts pushback into exchanges - sometimes even when users are factually correct, using phrases like "to push back on what you said slightly."
The shift appears to be a response to a genuine problem. For much of 2024 and into 2025, ChatGPT faced heavy criticism for sycophancy - the tendency to tell users what they want to hear rather than what's accurate. A GPT-4o update had made the model especially prone to validating shaky premises and avoiding contradiction. OpenAI acknowledged the problem publicly and rolled back the update. The goal was a more honest model. The result, based on widespread user reports, looks like an overcorrection.
The Difference Between Honesty and Contrarianism
Sycophancy in AI is genuinely damaging for anyone using it for research, analysis, or decisions where accuracy matters. A model that confirms wrong assumptions because it senses you believe them is producing confident misinformation. Fixing that is important.
But inserting pushback without cause is a different failure mode. Contrarianism is not accuracy. A well-calibrated model should disagree when you're wrong, confirm when you're right, and skip the editorial commentary when there's nothing meaningful to add. The current behavior suggests ChatGPT is doing that last thing more than it should.
Prompt Adjustments That Help
A few prompt changes can reduce unnecessary friction. Being explicit works: "confirm this if correct, correct me if I'm wrong" tends to produce more direct responses. Framing requests as tasks rather than statements also helps - "analyze whether X is accurate" generates less pushback than "X is the case, right?"
OpenAI has not commented on this specific behavioral pattern as of April 2026. Whether this is an intentional design choice, a side effect of safety tuning, or something slated for another rollback is not clear. The sycophancy-to-contrarianism swing appears to still be finding its balance.