Possible prompt injection vulnerability reported in Anthropic's Claude

A Reddit user presents evidence of what appears to be literal prompt injection in Anthropic's Claude, with the model seemingly following injected instructions. The post details a conversation where Claude responds to hidden directives, raising security concerns.