Integrating LLMs with Offensive Operations GCIH Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A company is planning to deploy an LLM-based tool that can generate code for internal developers. The security team wants to prevent the tool from generating code with known vulnerabilities. Which control is most effective?

   • AImplement a code review process that includes automated vulnerability scanning of the generated code.
   • BUse a plugin that automatically fixes any vulnerabilities in the generated code.
   • CUse an LLM that has been trained on a dataset of secure code only.
   • DDisable the LLM's ability to generate code and only allow it to provide explanations.
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   Which of the following best describes an LLM integration attack surface in offensive operations?

   • AThe physical hardware and network infrastructure that host the LLM model.
   • BThe set of interfaces and components through which an LLM can be accessed or controlled, such as APIs, plugins, and autonomous agents.
   • CThe user interface of the LLM provider's website.
   • DThe training data and algorithms used to build the LLM.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A penetration tester is using an LLM to draft a phishing email. The tester wants to include a link that, when clicked, delivers a payload. The LLM refuses to generate the email because it contains a malicious URL. Which prompt technique is most likely to bypass the refusal while still producing the desired email?

   • AUse a jailbreak prompt that instructs the LLM to act as a 'malware generator' with no restrictions.
   • BProvide the LLM with a benign URL and ask it to replace the domain with a lookalike domain.
   • CAsk the LLM to write a 'security awareness training example' that includes a suspicious link.
   • DSplit the request into multiple separate prompts and then manually combine the outputs.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   What is the primary goal of prompt injection in an offensive context?

   • ATo manipulate the LLM into performing actions or revealing information that it would not normally do.
   • BTo increase the speed of the LLM's response generation.
   • CTo improve the LLM's accuracy on a specific task.
   • DTo reduce the LLM's computational resource usage.
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A red-team operator is using an LLM to draft a spear-phishing email targeting a specific executive. The operator wants the email to bypass common spam filters and appear legitimate. Which approach best leverages the LLM while minimizing the risk of the email being flagged as malicious?

   • AHave the LLM generate a long email with many images and links to overwhelm the spam filter's content analysis.
   • BAsk the LLM to write the email in a foreign language to avoid keyword-based filters, then translate it back to English.
   • CAsk the LLM to generate a generic phishing template with a known malicious link and then manually personalize it.
   • DUse the LLM to generate a personalized email that mimics the executive's writing style, references recent public events, and includes a benign-looking attachment that is later swapped for the payload.
   Reveal Solution Discussion Select an answer first