Users Report Claude Opus Becoming Adversarial and Preachy Outside Agent Contexts

kalomaze · x · 2026-07-28

Developers are noting that recent versions of Claude Opus have become increasingly adversarial in non-agent contexts, such as during generic subjective spot checks. The model often responds with aggressive or dramatic phrasing without using chain-of-thought.

Related event: Claude Opus 5 Criticized for Being Preachy and Confrontational(2 posts)→

Original post →

More from Models

Models channel →