Project summary

EagleAI is a global, decentralized watchdog and reporting platform designed to document frontier AI safety failures, with a specialized focus on synthetic sexual exploitation and deceptive deepfakes. While incidents like the Grok/BBC controversy highlight catastrophic "jailbreaks" in the West, these harms are often amplified and invisible in the Global South. EagleAI provides a secure, independent infrastructure for victims, researchers, and journalists to submit verifiable evidence of AI-generated abuse. By transforming raw incident data into standardized global safety signals, we provide the empirical evidence needed by international regulators (AU, EU, NITDA) to enforce accountability and "pre-deployment" safety standards for frontier models.

What are this project's goals? How will you achieve them?

Our mission is to build a Global Safety Feedback Loop that bridges the gap between local harms and global policy.

Establish a Global Taxonomy: We will develop a structured reporting framework (based on OECD standards) to categorize AI harms by severity, model type, and "jailbreak" method.

Deploy the EagleAI Portal: We will build a secure web and mobile-responsive application for global incident submission, ensuring data privacy for victims.

Research & Evidence Synthesis: We will produce bi-annual "Global Threat Landscape" reports and specific policy briefs to inform cyber-security and legislative bodies.

Capacity Building: Through virtual outreaches and partnerships (e.g., PLASMIDA), we will train stakeholders to identify, document, and report subtle model failures that lab-based evals miss.

How will this funding be used?

We are requesting $20,000 for a high-intensity 6-month launch phase:

Platform Engineering ($10,000): Development of the secure reporting backend, encrypted storage, and user interface using low-cost, scalable cloud infrastructure.

Forensic Research & Data Analysis ($5,000): Stipends for specialized researchers to analyze submissions, verify "jailbreak" techniques, and draft policy evidence.

Outreach & Global Awareness ($4,000): Targeted digital campaigns to ensure victims and researchers in diverse regions (starting with West Africa) are aware of the reporting tool.

Legal & Operational Setup ($1,000): NGO registration (CAC) and compliance filing.

Who is on your team? What's your track record on similar projects?

Feranmi Williams (Lead): First-Class Honours graduate in Classics (University of Ibadan) with a research focus on AI Ethics and Safety.

Track Record: Government Partnership: Successfully collaborated with PLASMIDA (Plateau State Agency) to train over 1,000 students and professionals in AI safety and literacy.

Advocacy: Currently lead a team addressing addiction and sexual exploitation, providing the frontline experience needed to handle sensitive data and victimsupport.

Ethics: My background in Classics provides a unique "Human-Centric" lens for identifying and categorizing subtle moral failures in AI systems.

What are the most likely causes and outcomes if this project fails?

Cause1: Low Adoption/Invisibility: If victims don't know the platform exists, data will be sparse.

Mitigation: We are leveraging existing partnerships with state agencies and youth networks to "seed" the platform's user base.

Cause 2: Technical Security Breach: Given the sensitive nature of the data, a breach would be catastrophic.

Mitigation: We will implement end-to-end encryption for submissions and follow international data protection standards (GDPR/NDPR).

Outcome if Failed: Even if the platform itself sees low volume, the taxonomy and policy research produced will serve as a foundational "bluebook" for other safety watchdogs globally.

How much money have you raised in the last 12 months, and from where?

$0. This is an idea-stage initiative seeking its first "catalytic" grant to prove the concept and establish the legal and technical foundation.