Agentforce Guardrails combine user-defined safeguards and Salesforce managed protections to foster trust and maintain control over agent behaviour. Key mechanisms work in concert to prevent deviations from core instructions, block off-topic or inappropriate conversations and mitigate risks such as hallucination, bias or the dissemination of incorrect information.

The Einstein Trust Layer is a robust set of features and guardrails that protect the privacy and security of your data, improve the safety and accuracy of your AI results and promote the responsible use of AI across the Salesforce ecosystem. With features like dynamic grounding, zero data retention and toxicity detection, the Einstein Trust Layer is designed to help you unleash the power of generative AI without compromising your safety or security standards.

The Salesforce Platform is built on a foundation of trust, including security testing, site reliability, disaster recovery, least privileged access and core access controls embedded into our deeply unified platform. Our shared responsibility model allows customers to establish a robust governance foundation with built-in security controls while offering additional solutions to customise and strengthen data security, compliance and resilience, including products to encrypt, audit, monitor, mask, protect, manage, backup and archive data.