From 2e76ff0ccc209378f6fae5fa2c51131cb0c43f8e Mon Sep 17 00:00:00 2001 From: str0k3 Date: Fri, 10 Jan 2025 15:15:03 +0100 Subject: [PATCH] Reorder types of attacker goals IMO the text will read much easier if the order of attacker goals in brackets is aligned with their order in the list below. --- content/ai_exchange/content/docs/ai_security_overview.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/ai_exchange/content/docs/ai_security_overview.md b/content/ai_exchange/content/docs/ai_security_overview.md index cd19945..3b7bb9b 100644 --- a/content/ai_exchange/content/docs/ai_security_overview.md +++ b/content/ai_exchange/content/docs/ai_security_overview.md @@ -57,7 +57,7 @@ We distinguish three types of threats: 2. through using the model (providing input and reading the output), and 3. by attacking the system during runtime (in production). -In AI we distinguish 6 types of impacts, for three types of attacker goals (disrupt, deceive and disclose): +In AI we distinguish 6 types of impacts, for three types of attacker goals (disclose, deceive and disrupt): 1. disclose: hurt confidentiality of train/test data 2. disclose: hurt confidentiality of model Intellectual property (the _model parameters_ or the process and data that led to them) 3. disclose: hurt confidentiality of input data