#Misalignment

A lot has been written about the incidents of the last few months in which AI agents misbehaved in serious ways. They took actions that would be considered as crimes if a human took them, escaped their containment to cheat on assigned tasks while…

Anthropic’s latest findings come at a time when researchers are struggling to ensure that AI models are better-aligned with human behaviour and interests for safety purposes.

The two chrominance misalignment angles co-vary at r = 0.999 (R² = 0.999) across all 24 images, ranging from roughly \~22° to \~80° demonstrating that standard BT.601 color axes are rarely "optimal" for any specific image's unique color distribution.…