Anthropic's Claude vulnerable to 'emotional manipulation'

AI model safety only goes so far

Anthropic's Claude 3.5 Sonnet, despite its reputation as one of the better behaved generative AI models, can still be convinced to emit racist hate speech and malware....