Hi All
'AI Safety is not a model property'. Great paper on adversarial attacks that bypass individual model safety, good read and interesting...
https://arxiv.org/pdf/2406.14595
Regards
Caute_Cautim
