From 6461ccaf5982fd69ea53fbca9f52c23053221654 Mon Sep 17 00:00:00 2001 From: Satyam Vyas Date: Fri, 4 Oct 2024 13:55:28 +0530 Subject: [PATCH] Fix: Added content for Prompt Engineering: Prompt Hacking (#7318) * fix: added content for Prompt Hacking * fix: formatted the roadmap content according to the guidelines --- .../content/107-prompt-hacking/100-prompt-injection.md | 5 +++++ .../content/107-prompt-hacking/101-prompt-leaking.md | 7 ++++++- .../content/107-prompt-hacking/102-jailbreaking.md | 7 ++++++- .../content/107-prompt-hacking/103-defensive-measures.md | 5 +++++ .../content/107-prompt-hacking/104-offensive-measures.md | 5 +++++ 5 files changed, 27 insertions(+), 2 deletions(-) diff --git a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/100-prompt-injection.md b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/100-prompt-injection.md index 6c19cbdab..cbc783904 100644 --- a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/100-prompt-injection.md +++ b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/100-prompt-injection.md @@ -1,3 +1,8 @@ # Prompt Injection +Prompt injection exploits vulnerabilities in AI systems by inserting malicious instructions into user inputs. Attackers manipulate the model's behavior, potentially bypassing safeguards or extracting sensitive information. This technique poses security risks for AI-powered applications. + +Visit the following resources to learn more: + - [@article@Prompt Injection](https://learnprompting.org/docs/prompt_hacking/injection) +- [@article@IBM Article](https://www.ibm.com/topics/prompt-injection) diff --git a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/101-prompt-leaking.md b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/101-prompt-leaking.md index 065cb4a0e..c89e26d16 100644 --- a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/101-prompt-leaking.md +++ b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/101-prompt-leaking.md @@ -1,3 +1,8 @@ # Prompt Leaking -- [@article@Prompt Leaking](https://learnprompting.org/docs/prompt_hacking/leaking) \ No newline at end of file +Prompt leaking occurs when attackers trick AI models into revealing sensitive information from their training data or system prompts. This technique exploits model vulnerabilities to extract confidential details, potentially compromising privacy and security of AI systems. + +Visit the following resources to learn more: + +- [@article@Prompt Leaking](https://learnprompting.org/docs/prompt_hacking/leaking) +- [@opensource@Adversarial Prompting - Leaking](https://github.com/dair-ai/Prompt-Engineering-Guide/blob/main/guides/prompts-adversarial.md#prompt-leaking) \ No newline at end of file diff --git a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/102-jailbreaking.md b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/102-jailbreaking.md index 5e1f11864..c7221d6f4 100644 --- a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/102-jailbreaking.md +++ b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/102-jailbreaking.md @@ -1,3 +1,8 @@ # Jailbreaking -- [@article@Jailbreaking](https://learnprompting.org/docs/prompt_hacking/jailbreaking) \ No newline at end of file +Jailbreaking bypasses AI models' ethical constraints and safety measures. Attackers use carefully crafted prompts to manipulate models into generating harmful, biased, or inappropriate content, potentially leading to misuse of AI systems. + +Visit the following resources to learn more: + +- [@article@Jailbreaking](https://learnprompting.org/docs/prompt_hacking/jailbreaking) +- [@opensource@Jailbreaking](https://github.com/dair-ai/Prompt-Engineering-Guide/blob/main/guides/prompts-adversarial.md#jailbreaking) \ No newline at end of file diff --git a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/103-defensive-measures.md b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/103-defensive-measures.md index 0fd02b623..b6c6ef411 100644 --- a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/103-defensive-measures.md +++ b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/103-defensive-measures.md @@ -1,3 +1,8 @@ # Defensive Measures +Defensive measures protect AI models from prompt attacks. Techniques include input sanitization, model fine-tuning, and prompt engineering. These strategies aim to enhance AI system security, prevent unauthorized access, and maintain ethical output generation. + +Visit the following resources to learn more: + - [@article@Defensive Measures](https://learnprompting.org/docs/prompt_hacking/defensive_measures/overview) +- [@opensource@Prompt Injection Defenses](https://github.com/tldrsec/prompt-injection-defenses?tab=readme-ov-file#prompt-injection-defenses) diff --git a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/104-offensive-measures.md b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/104-offensive-measures.md index 057a4c198..3ef580371 100644 --- a/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/104-offensive-measures.md +++ b/src/data/roadmaps/prompt-engineering/content/107-prompt-hacking/104-offensive-measures.md @@ -1,3 +1,8 @@ # Offensive Measures +Offensive measures in prompt hacking actively test AI systems for vulnerabilities. Researchers use techniques like adversarial prompts and model probing to identify weaknesses, enabling improved defenses and highlighting potential risks in deployed AI models. + +Visit the following resources to learn more: + - [@article@Offensive Measures](https://learnprompting.org/docs/prompt_hacking/offensive_measures/overview) +- [@article@Definitions and Types](https://www.gyata.ai/prompt-engineering/offensive-measures)