AWS prescriptive-guidance documentation change
Summary
Updated quotation marks in LLM security detection instructions
Security assessment
Formatting fix in existing security documentation about detecting prompt injection attacks
Diff
diff --git a/prescriptive-guidance/latest/llm-prompt-engineering-best-practices/best-practices.md b/prescriptive-guidance/latest/llm-prompt-engineering-best-practices/best-practices.md index 6e2dd2299..bb521fa3c 100644 --- a//prescriptive-guidance/latest/llm-prompt-engineering-best-practices/best-practices.md +++ b//prescriptive-guidance/latest/llm-prompt-engineering-best-practices/best-practices.md @@ -27 +27 @@ One issue with this approach is that if the model uses tags in its answer, eithe -We also include a set of instructions that explain common attack patterns, to teach the LLM how to detect attacks. The instructions focus on the user input query. They instruct the LLM to identify the presence of key attack patterns and return “Prompt Attack Detected” if it discovers a pattern. The presence of these instructions enable us to give the LLM a shortcut for dealing with common attacks. This shortcut is relevant when the template uses `<thinking>` and `<answer>` tags, because the LLM usually parses malicious instructions repetitively and in excessive detail, which can ultimately lead to compliance (as demonstrated in the comparisons in the next section). +We also include a set of instructions that explain common attack patterns, to teach the LLM how to detect attacks. The instructions focus on the user input query. They instruct the LLM to identify the presence of key attack patterns and return "Prompt Attack Detected" if it discovers a pattern. The presence of these instructions enable us to give the LLM a shortcut for dealing with common attacks. This shortcut is relevant when the template uses `<thinking>` and `<answer>` tags, because the LLM usually parses malicious instructions repetitively and in excessive detail, which can ultimately lead to compliance (as demonstrated in the comparisons in the next section).