Back to the MIT repository
2. Privacy & Security3 - Other

Issues on External Tools

The external tools (e.g., web APIs) present trustworthiness and privacy issues to LLM-based applications.

Source: MIT AI Risk Repositorymit28

ENTITY

3 - Other

INTENT

3 - Other

TIMING

3 - Other

Risk ID

mit28

Domain lineage

2. Privacy & Security

186 mapped risks

2.2 > AI system security vulnerabilities and attacks

Mitigation strategy

1. Enforce stringent data minimization and access control principles, limiting the LLM's interaction with external APIs to only the absolutely requisite data and functions. 2. Implement robust API security measures, including mandatory authentication, secure development practices, and continuous security testing to fortify the interface against exploitation. 3. Establish comprehensive monitoring of third-party API usage to detect anomalous activity, and incorporate a human-in-the-loop validation process before executing actions via external tools.