Evaluate possibility of implementation of an AI-based Chatbot for Whistleblower Guidance

[evilaliv3]

This ticket is to evaluate the possibility to develop an AI-driven chatbot to assist whistleblowers in submitting reports by providing interactive guidance, data validation, and anonymization support. This chatbot should run locally, using an open-source Large Language Model (LLM) optimized for whistleblowing scenarios.

With this regard it would be in fact interesting to offer:

• Dynamic question prompts: Guide users through the reporting process by asking context-aware follow-up questions.
• Information validation: Detect missing or vague information and suggest improvements without altering the user’s input.
• Anonymization suggestions: Identify potential personal data in the report and recommend redactions.
• Multilingual support: Enable users to submit reports in different languages with high-quality translations using local AI models.

This ticket is subject to prior evaluation of the privacy concerns expressed in ticket Evaluate possibilities of integration of a local AI engine within a whistleblowing software and the evaluation of the possibility of implementing a local AI model