Readplace

'Claude cannot be trusted to perform complex engineering tasks': AMD AI head slams Anthropic's coding…

TechRadar 3 min read
View original
Summary (TL;DR)
AMD AI director Stella Laurenzo says Claude Code lost performance after a February 2026 update, based on analysis of over 6,800 coding sessions. Stop-hook violations increased from zero to roughly 10 per day. Laurenzo links the decline to thinking redaction, shift from research-first to edit-first behavior, and reduced code quality. Anthropic responded that redaction only hides reasoning, and introduced adaptive thinking. They recommend setting effort=high for complex tasks, and plan to test higher effort for Teams and Enterprise users.