Prompt defense system policies

Key Takeaway

Prompt defense uses system policies, which are instructions that tell the large language model (LLM) how to behave, to lower the chance of unintended or harmful output.

Prompt defense is one of the protections in the Einstein Trust Layer. It shapes how the LLM behaves, and it helps defend against jailbreaking and prompt injection attacks.

Prompt defense in the Einstein Trust Layer

Einstein Trust Layer capabilities, such as prompt defense, apply only to generative AI and Agentforce features. Prompt defense is available in:

Prompt Builder
Prompt Template Connect API
Prompt Template Invocable Action

Agentforce Specialist Study Guide

Checklist progressProgress

0/204Learned

Checklist progressProgress

0/204Learned

Given business requirements, identify when it's appropriate to use Prompt Builder.

0/5

Identify access controls governing prompt templates

0/5

Identify the considerations for using a prompt template type such as field generation and flex types.

0/4

Given a scenario, identify the appropriate grounding technique.

0/11

Explain the process for creating, activating, and executing prompt templates.

0/5

Explain how to implement best practices for writing effective prompts.

0/10

Identify the security and privacy features of the Trust Layer

0/8

Einstein Trust Layer architecture

Learn this concept

Secure data retrieval for prompt grounding

Learn this concept

Pattern-based and field-based LLM data masking

Learn this concept

Prompt defense system policies

Learn this concept

Zero-data retention with third-party LLMs

Learn this concept

Toxicity detection on generated responses

Learn this concept

Generative AI audit trail and feedback data

Learn this concept

LLM data masking for agents versus embedded generative AI features

Learn this concept

Explain how to manage and prevent specific models from being accessed

0/5

Prepare for the Exam

Play Today's Certle

Study Community

Ask questions and get the latest info from other Agentforce Specialist studiers.

Go to Discord

Prompt defense system policies

Key Takeaway

Prompt defense uses system policies, which are instructions that tell the large language model (LLM) how to behave, to lower the chance of unintended or harmful output.

Prompt defense is one of the protections in the Einstein Trust Layer. It shapes how the LLM behaves, and it helps defend against jailbreaking and prompt injection attacks.

Prompt defense in the Einstein Trust Layer

Einstein Trust Layer capabilities, such as prompt defense, apply only to generative AI and Agentforce features. Prompt defense is available in:

Prompt Builder
Prompt Template Connect API
Prompt Template Invocable Action