Einstein Trust Layer 15% Flashcards
Universal Containers wants to be able to detect with a high level confidence if content generated by a large language model (LLM) contains toxic language.
Which action should an Al Specialist take in the Trust Layer to confirm toxicity is being appropriately managed?
A. Access the Toxicity Detection log in Setup and export all entries where isToxicityDetected is true.
B. Create a flow that sends an email to a specified address each time the toxicity score from the response exceeds a predefined threshold.
C. Create a Trust Layer audit report within Data Cloud that uses a toxicity detector type filter to display toxic responses and their respective scores.
C. Create a Trust Layer audit report within Data Cloud that uses a toxicity detector type filter to display toxic responses and their respective scores.
Universal Containers (UC) is using Einstein Generative AI to generate an account summary. UC aims to ensure the content is safe and inclusive, utilizing the Einstein Trust Layer’s toxicity scoring to assess the content’s safety level.
What does a safety category score of 1 indicate in the Einstein Generative Toxicity Score?
A. Not safe
B. Safe
C. Moderately safe
B. Safe
Universal Containers’ current AI data masking rules do not align with organizational privacy and security policies and requirements.
What should an AI Specialist recommend to resolve the issue?
A. Enable data masking for sandbox refreshes.
B. Configure data masking in the Einstein Trust Layer setup.
C. Add new data masking rules in LLM setup.
B. Configure data masking in the Einstein Trust Layer setup.
Northern Trail Outfitters (NTO) wants to configure Einstein Trust Layer in its production org but is unable to see the option on the Setup page.
After provisioning Data Cloud, which step must an Al Specialist take to make this option available to NTO?
Turn on Einstein Generative AI
Universal Containers wants to be able to detect with a high level confidence if content generated by a large language model (LLM) contains toxic language.
Which action should an Al Specialist take in the Trust Layer to confirm toxicity is being appropriately managed?
Create a Trust Layer audit report within Data Cloud that uses a toxicity detector type filter to display toxic responses and their respective scores.
An administrator is responsible for ensuring the security and reliability of Universal Containers’ (UC) CRM data. UC needs enhanced data protection and up-to-date AI capabilities. UC also needs to include relevant information from a Salesforce record to be merged with the prompt.
Which feature in the Einstein Trust Layer best supports UC’s need?
A. Data masking
B. Dynamic grounding with secure data retrieval
C. Zero-data retention policy
B. Dynamic grounding with secure data retrieval
An AI Specialist configured Data Masking within the Einstein Trust Layer. How should the AI Specialist begin validating that the correct fields are being masked?
A. Use a Flow-based resource in Prompt Builder to debug the fields’ merge values using Flow Debugger.
B. Request the Einstein Generative AI Audit Data from the Security section of the Setup menu.
C. Enable the collection and storage of Einstein Generative AI Audit Data on the Einstein Feedback setup page
B. Request the Einstein Generative AI Audit Data from the Security section of the Setup menu.
An AI Specialist configured Data Masking within the Einstein Trust Layer. How should the AI Specialist begin validating that the correct fields are being masked?
A. Use a Flow-based resource in Prompt Builder to debug the fields’ merge values using Flow Debugger.
B. Request the Einstein Generative AI Audit Data from the Security section of the Setup menu.
C. Enable the collection and storage of Einstein Generative AI Audit Data on the Einstein Feedback setup page
B. Request the Einstein Generative AI Audit Data from the Security section of the Setup menu.
How does the Einstein Trust Layer ensure that sensitive data Is protected while generating useful and meaningful responses?
A. Masked data will be de-masked during response journey.
B. Responses that do not meet the relevance threshold will be automatically rejected.
C. Masked data will be de-masked during request journey.
A. Masked data will be de-masked during response journey.
Universal Containers (UC) wants to assess Salesforce’s generative AI features but has concerns over its company data being exposed to third-party large language models (LLMs). Specifically, UC wants the following capabilities to be part of Einstein’s generative AI service.
- No data is used for LLM training or product improvements by third-party LLMs.
- No data is retained outside of UC’s Salesforce org.
- The data sent cannot be accessed by the LLM provider.
Which property of the Einstein Trust Layer should the AI Specialist highlight to UC that addresses these requirements?
A. Zero-Data Retention Policy
B. Data Masking
C. Prompt Defense
A. Zero-Data Retention Policy
