The Einstein GPT Trust Layer elevates the security of generative AI through data and privacy controls that are seamless to the end user. This uses a number of features to ensure that Generative AI into Salesforce is safe and secure.
What is Einstein GPT Trust Layer?
Salesforce GPT Trust Layer is the solution to bridging that AI trust gap. The Einstein GPT Trust Layer enables to benefit from generative AI without compromising on data security and privacy controls. It uses a technique called dynamic grounding to take that customer context in your CRM and use it to safely enrich the prompts that are delivered to the LLMs.
Learn more about Salesforce and ChatGPT use cases and Integration.
How Einstein GPT Trust Layer works?
Here some pillar of GPT Trust Layer. In the CRM App from the app you are doing prompt. The prompt will combine with your company data and secure data retrial. Salesforce do the dynamic grounding and do data masking if needed. Then it go to Salesforce Secure gateway. In gateway it allow to talk to the models. Salesforce make sure non of data stored in Salesforce.
1. Secure Data Retrial
With Secure Data Retrial you can pull data from Salesforce using Data Cloud. This enables more relevant outputs from those models to build contextual prompts. Secure Data Retrial focuses on ensuring the security of data used by the Generative AI models.
2. Dynamic Grounding
Dynamic grounding drive an LLM’s answers using the correct information, grounding the model in factual data and relevant context. Using this helps prevent the generation of irrelevant or misleading information by enhancing the accuracy and relevance of the AI-generated responses.
3. Data Masking
Data Masking is important layer for data privacy. With the GPT Trust Layer, Salesforce can strip all PII data from prompts using data masking tools so that sensitive data is even processed by the LLM.
4. Toxicity Detection
Any generation from those models is scanned for toxicity and bias. It will help to flagging the date speech and negative prompts.
5. Auditing
Trust layer will audited for compliance reasons and make care you get high quality data.
6. Zero Retention
Most importantly, your data will never stored outside of Salesforce. Once model processes your prompt, the prompt and the generation are both forgotten. Data will not stored for any monitoring, not for quality, and certainly not for data training.
Learn more.
Einstein GPT Trust Layer Video By Salesforce
Summary
Trust is core principal for Salesforce. Learn more about Salesforce Einstein GPT.
