Announcement
RT-SAFE's reported results: 94.1% reach the goal; 0.7% finish without a safety event
A post introduces RT-SAFE as a way to evaluate embodied AI safety while the world keeps moving.
TLDR
A post introducing RT-SAFE says it evaluates safety for embodied AI in a changing world. For eight frontier vision-language models, the post says 94.1% reach the goal, but only 0.7% finish without a safety event.
Combined views
875
2 Sources, first seen ago
