Why LinkedIn has a trust agent

Key Concepts

• Trust Agent: An internal tool developed at LinkedIn to identify and mitigate potential trust vulnerabilities and harm vectors in new features or product specifications.
• Harm Vectors: Potential pathways or methods through which a feature or product could be exploited or cause harm to users.
• Open to Work Feature: A LinkedIn feature allowing job seekers to signal their availability, which, while beneficial, also introduces trust risks.
• Vulnerability Assessment: The process of identifying weaknesses in a system or feature that could be exploited.
• Proactive Risk Mitigation: Identifying and addressing potential harms before they occur.

Trust Agent at LinkedIn

LinkedIn developed its own internal "trust agent" to address the unique trust vectors inherent in its platform. This agent is designed to incorporate LinkedIn's specific know-how, context, and information base.

Functionality of the Trust Agent

When a new feature or idea is being developed, its specification is run through the trust agent. The agent then identifies potential vulnerabilities and harm vectors that might be introduced by the proposed feature. This process allows for proactive identification of risks.

Case Study: "Open to Work" Feature

The "Open to Work" feature, designed to help job seekers signal their availability, serves as a prime example of the trust agent's utility.

• Benefit: The feature is a valuable signal for job seekers and has shown significant success.
• Trust Vector Introduced: It creates a trust vector for bad actors because individuals actively looking for jobs are potentially more vulnerable to scams.
• Trust Agent Application: The specification for the "Open to Work" feature was run through the trust agent.
• Outcomes: The agent not only identified the risks that the LinkedIn team had already anticipated and initiated mitigation for, but also uncovered additional vulnerabilities that were not caught until later stages of development. This highlights the agent's ability to find unforeseen issues.

Key Arguments and Perspectives

The core argument presented is that proactive identification and mitigation of trust vulnerabilities are crucial for platform safety, especially for features that target potentially vulnerable user groups. The trust agent is presented as an effective tool for achieving this.

Conclusion

The trust agent at LinkedIn is a critical internal tool that helps to systematically identify and address trust vulnerabilities and harm vectors in new product features. The "Open to Work" feature exemplifies how this agent can uncover risks that might otherwise be missed, leading to more secure and trustworthy product development. The emphasis is on bringing context-specific know-how into an automated assessment process to prevent potential harm to users.