Get the latest tech news
Data Exfiltration from Slack AI via indirect prompt injection
Authors: PromptArmor
This vulnerability can allow attackers to steal anything a user puts in a private Slack channel by manipulating the language model used for content generation. After August 14th, Slack also ingests uploaded documents, Google Drive files, etc which increases the risk surface area as we’ll address in section 3. If a user downloads a PDF that has one of these malicious instructions (e.g. hidden in white text) and subsequently uploads it to Slack, the same downstream effects of the attack chain can be achieved.
Or read this on Hacker News
