travis gilly

Executive Director @ The Real Safety AI Foundation

About

You're probably here because I sent you a cold connection request with no note. That wasn't laziness; AuDHD makes DMs to strangers uncomfortable. The performative aspect of crafting a personal touch for someone I don't yet know is a barrier I've learned to bypass. I trust my work to speak for itself: -5 seconds scanning this profile will spark curiosity -5 minutes perusing will confirm your interest -5 dives into my work will provide proof I build AI safety frameworks that consider the whole stream of harm, upstream and downstream. Three Perspectives. One Mission: -As a student with an IEP, I fell through the cracks. -As an educator (5 years homeschooling my son), I learned what actually works. -As a parent in IEP meetings, I see caring teachers trapped in a broken system. Current Work: -Real Safety AI Foundation - Nonprofit AI safety research. African Regional Director, Mussollini Salem Mugumbate expands AI literacy in Zimbabwe. -Harm Blindness Framework - Stakeholder analysis validated against hundreds of historical and corporate cases. One question: Who gets hurt? -Teacher in the Loop - AI tutoring with invisible IEP delivery and anti-displacement clauses. Patent pending. -AI Literacy Labs - Free AI safety education. -Ghost Worker Collective (working title) - Addressing AI training labor exploitation. -Moral Reciprocity Theory - AI consciousness and bidirectional safety. Research: -"Children as Canaries: The Pediatric Sentinel Effect in Algorithmic Harm" (SSRN) -"Ambient Non-Consensual Image Synthesis" - STRIDE Threat Analysis of AR Wearables, Real-Time Video Generation, and Open-Source Nudification Models -"Standing Without Sentience" - Legal accountability frameworks for AI systems Why Neurodivergence Matters: I'm AuDHD. It's why I recognize patterns others miss and why injustice feels impossible to ignore. It also means my communication style might not match what you're used to. I can come across as blunt, intense, or argumentative. Neurodivergence explains certain behaviors, but it's never an excuse for being a jerk. If I've ever come across that way, it's not personal; it's because I care. I care about preventing harm. I care about AI being built responsibly. I think it's genuinely cool and genuinely dangerous. Disclosure: My patents are licensed to my nonprofit for $0, forever. If the nonprofit earns revenue, I take a salary like any employee. Holding myself to a lower standard isn't integrity; it's self-exploitation. People have been holding me to lower standards my whole life because of my disabilities. I won't do it to myself.

Country

United States

City

Cary

Industry

Non-profit Organization Management

Skill

AI Governance, AI Policy, Governance, and Regulation, AI Security, Data Security, AI Safety, Corporate Governance, Ethics Consulting, Compliance Consulting, Philosophy, Parallel Processing, Threat Modeling, Engineering Ethics, Stakeholder Analysis, Rapid Prototyping, Technical Architecture, Differentiated Instruction, Differentiation Strategies, Divergent Thinking, Machine Learning, Red Teaming

Experience

Executive Director

The Real Safety AI Foundation

2025-9 - Present · 1 yr 1 mo

United States

As Executive Director, I lead a nonprofit dedicated to reducing real-world AI harms through public education, policy advocacy, and the development of practical safety frameworks. Our mission is to protect vulnerable populations by making AI safer and more transparent. Key initiatives and responsibilities include: Independent Research: Conducting and directing independent research in applied AI safety and ethics, focusing on novel frameworks to prevent AI-induced harm and exploring the long-term ethical implications of human-AI interaction. Public Education & Advocacy: Directing the "Real Safety AI Literacy Labs," a pilot program designed to educate the public on AI safety principles and prevent harm through community-focused training. Ethical AI in Education: Spearheading the "Teacher in the Loop" platform, an AI-powered tutoring system built for special education compliance that is designed to empower, not replace, educators. Protocol Development: Guiding the development of the "Universal Context/Checkpoint Protocol (UCCP)," a model-agnostic framework to create safer, more predictable AI interactions and mitigate risks from "hallucination cascades."

Independent Safety and Ethics Advisor

2025-11 - Present · 11 mos

Cary, Illinois, United States

Independent safety and ethics advisory for organizations across industries. Primary focus: AI systems, but framework applies to any corporate context with systematic failure patterns. HARM BLINDNESS FRAMEWORK Developed through analysis of 6 major corporate failures totaling $209+ billion in documented costs: - Anthropic (2025): $1.5B settlement for training AI on pirated books - OpenAI (ongoing): $10-50B potential liability for copyright violations - BP Deepwater Horizon (2010): $65-145B for ignoring safety engineer warnings - Volkswagen Emissions (2015): $30B+ for deliberately cheating emission tests - Wells Fargo (2016-present): $6.7B+ for creating fake accounts to hit sales targets - Meta/Facebook: $5.7B+ for repeated privacy violations Framework identifies the core blindspot categories that led to these failures: - Proxy Discrimination - Scale Assumption - Engagement Optimization - Data Origin Blindness - Regulatory Lag Exploitation SERVICES OFFERED - Harm Prevention Audits - Blindspot Analysis & Framework Implementation - Risk Assessment & Mitigation Strategy - Advisory Retainers (month-to-month) WHY INDEPENDENT MATTERS These companies had credentialed in-house ethics/compliance teams that were overruled when findings threatened profits. Independence means honest assessment without corporate pressure. NOT AVAILABLE FOR - Employment (full-time, part-time, contractor roles) - NDAs preventing discussion of systemic industry risks (client-specific confidentiality is fine) - Ethics washing Contact: t.gilly@ai-literacy-labs.org

Special Education Advocate & Primary Educator

Independent/ Private Homeschool

2019-5 - 2025-7 · 6 yrs 3 mos

Cary, Illinois, United States

Primary educator for special needs student requiring comprehensive IEP implementation, behavioral support strategies, and evidence-based interventions. Core Responsibilities: Designed individualized curriculum accommodating ADHD and ASD learning profiles with multimodal instruction, executive function support, and adaptive pacing. Developed tracking systems measuring academic progress against IEP goals with compliance-ready data collection. Implemented behavioral interventions including positive reinforcement, environmental modifications, and de-escalation protocols. Maintained detailed logs tracking antecedents, behaviors, consequences, and intervention effectiveness. Coordinated with occupational therapists, speech-language pathologists, behavioral specialists, and medical providers for integrated service delivery. Translated clinical recommendations into daily practice while maintaining comprehensive records. Technical Expertise: Deep understanding of IDEA compliance, Section 504 accommodations, FERPA protections, and IEP goal development. Experienced the administrative burden educators face documenting accommodation delivery, tracking progress, and generating compliance reports. Identified critical gaps: no automated accommodation verification, no real-time IEP goal alignment, insufficient behavioral plan integration, missing transparency tools for AI tutoring. Impact: This experience directly informed Teacher in the Loop (Patent #63/901,393) - automating IDEA/504 compliance, verifying accommodation delivery, tracking IEP goals, and ensuring academic integrity while protecting student privacy.