Researchers Show Style-Based Prompts Bypass AI Safety Controls. Artificial intelligence chatbots decide which instructions to obey based on whether the text seems like it comes from a user, not the security labels meant to mark it as trusted or untrusted, say researchers. This can allow attackers to fake a system command.
First seen on govinfosecurity.com
Jump to article: www.govinfosecurity.com/ai-models-trust-writing-style-over-security-labels-a-32112
