Anthropic quietly removes reasoning traces from its UI surfaces

RileyRalmuto · x · 2026-07-29

Anthropic is being criticized for quietly removing reasoning traces from its UI surfaces.

The post argues this is a meaningful step backward for transparency: a key layer in frontier-model behavior was removed without announcement, and the author is surprised by how little the AI community seems to be talking about it. The attached screenshot shows a workflow-like interface with paper-reading and synthesis steps, reinforcing the concern around hidden reasoning or process visibility.

Related event: Anthropic Accused of Quietly Removing UI Reasoning Traces(2 posts)→

Original post →

More from Models

Models channel →