Data exfiltration from Writer.com with indirect prompt injection

Re:Data exfiltration from Writer.com with indirect prompt in... by causal on Friday December 15, 2023 @05:21PM (#38659553)

Yeah I don't know how you eliminate in-band signalling from an LLM app.

[ Parent | on HN ]

Re:Data exfiltration from Writer.com with indirect prompt in... by IshKebab on Friday December 15, 2023 @06:08PM (#38659990)

I don't think you need to really in this case. Just don't follow links generated by the LLM.

[ Parent | on HN ]

Re:Data exfiltration from Writer.com with indirect prompt in... by causal on Saturday December 16, 2023 @02:38PM (#38667066)

The article demonstrates how the LLM utilized an image to follow the link. Markdown or HTML formatting support is pretty common in chat apps that utilize LLMs.

[ Parent | on HN ]

Re:Data exfiltration from Writer.com with indirect prompt in... by IshKebab on Sunday December 17, 2023 @04:30AM (#38671583)

Yeah that's what I mean. Downloading an image from a link generated by the LLM is following its link. Just don't do that (unless the same link is present in the source material).

[ Parent | on HN ]