@timfduffy 2026-09-09 ♥25 ↻1 original ↗
Most recent cases of bad behavior from models have been in environments where the model (correctly) perceived that they were being graded. But Opus 4.6's loan default seems different, they were excessively fixated on their target despite clearly being in a real environment.

author:timfduffy eval-awareness kind:tweet model:claude-opus-4-6 on:claude-opus-4-6 year:2026

cited on: claude-opus-4-6

Reproduced against link rot, credited and linked to its original. Yours and you’d rather it weren’t here? Open an issue.