Introduction

Dalil is developed and owned by Siren Analytics. At Siren Analytics, we are committed to building and deploying AI technologies that serve the public good and protect fundamental human rights. Dalil, our AI-powered disinformation detection system, is designed to support truth, accountability, and information integrity across digital platforms. This document outlines the principles, safeguards, and ethical commitments that guide the development and use of our AI technology. 

Purpose and scope

This Responsible AI statement applies to all components of our disinformation detection platform, including:

• Machine learning models (e.g. credibility scoring, clustering, threat detection, etc...)

• Human-in-the-loop workflows (e.g. analyst validation, lead confirmation, fact-checking, ...)

• Data collection and labeling

• Model evaluation and deployment
  It also governs internal processes, third-party integrations, and external use cases of the system.

Core ethical principles

Fairness and Non-Discrimination

We ensure that our AI models are trained, tested, and monitored to reduce biases across race, gender, ethnicity, geography, political orientation, and language. Disinformation impacts people unequally; our system strives to mitigate such disparities without amplifying them.

Privacy and Data Protection

User and source data are handled in compliance with applicable data protection laws (e.g. GDPR, LGPD). We implement strong security protocols and only collect minimal, necessary data. Our system does not profile individuals or store identifiable user data without consent; except standard usage related data for billing and performance purposes.

Transparency and Explainability

We prioritize interpretability by providing:

• Scoring rationales

• Justifications for clustering decisions

• Audit trails for any flagged content
  We acknowledge AI limitations and communicate clearly when results are probabilistic or AI-assisted.

Human Oversight and Accountability

All major decisions — including the classification of a threat or campaign — involve human analysts/leads. AI augments, but does not replace, expert judgment. We assign clear roles and responsibilities for oversight, including:

• Ethical AI lead

• Product team accountability

• Incident reporting mechanisms

Safety and Harm Prevention

We proactively assess our models for the potential to cause direct or indirect harm (e.g., wrongful labeling, censorship risks). We reject cases that may suppress legitimate free speech or be used for surveillance purposes against vulnerable groups.

Integrity Against Misuse

We implement safeguards to:

• Prevent adversarial attacks (e.g., poisoning, evasion)

• Restrict misuse by authoritarian regimes, troll farms, or commercial disinformation actors

• Vet partners and clients through ethical reviews

Model Governance & Lifecycle 

The following outlines the ethical considerations throughout the AI lifecycle: 

Stakeholder Engagement

We actively seek input from:

• Journalists, researchers, and fact-checkers

• Civil society and human rights organizations

• Policy experts and regulators
  This helps ensure our system remains aligned with evolving ethical standards and real-world needs.

Commitments to Continuous Improvement

• Public disclosure of key findings or updates when risks are identified

• Dedicated feedback channel for users, watchdogs, and researchers

Contact and Feedback

For inquiries, partnerships, or reporting ethical concerns, please contact: hello@dalil.io