LLM Prompt Crafting

Type: technique

Description: The adversary uses their acquired knowledge of the target AI system to craft prompts that bypass its defenses. The adversary may iterate on the prompt to ensure that it works as-intended consistently.

Version: 0.1.0

Created At: 2025-03-04 10:27:40 -0500

Last Modified At: 2025-03-04 10:27:40 -0500

External References

Related Objects

• --> Resource Development (tactic): An adversary can craft a prompt that would circumvent the target AI system defenses.
• --> Commercial License Abuse (technique): For commercial products, prompt crafting can be easier to performed on an attacker-controlled tenant.
• --> LLM Jailbreak (technique): Prompt crafting typically involves jailbreaking.
• --> LLM Prompt Injection (technique): Prompt crafting typically involves prompt injection.
• <-- Microsoft Copilot Purview Audit Log Evasion and DLP Bypass (procedure): Using prior knowledge on how Copilot renders references.

Related Frameworks