AI Research x Governance

Exploring how frontier AI systems shape language, risk, and society.

Testing the limits of AI.

Listen to Madison’s recent interview with Sarah Senk (Cal Poly) and Taiyo Inoue (Cal State) at the California Learning Lab where we discuss the limits of AI translation, what AI safety practices like red-teaming reveal about model guardrails, and why human oversight remains essential.

About

Madison Van Doren is an AI researcher and legal scholar focused on the development, deployment, and regulation of frontier systems. Her work has been presented at venues including ACL, NeurIPS, AAAI, and the Linguistic Society of America, including release of FROMAGE, an open-source benchmark designed to assess cultural nuance in machine translation.

She was awarded the U.S. State Department NSLI-Y scholarship to study in Seoul, South Korea and graduated magna cum laude from Colorado State University with a degree in Linguistics. Madison is currently pursuing her juris doctorate at the University of Chicago where she focuses on AI governance, intellectual property, and the rapidly evolving legal landscape of emerging technology.

Passionate about global AI innovation and community-building, Madison advocates for sustainable and equitable growth in the industry through her research and philanthropy. She volunteers her time as a professional mentor for emerging female talent in Ethiopia with the Na’amal Foundation and hosts a monthly Women in AI happy hour in San Francisco.

Publications

  • [ACL 2026] This study expands prior work on multilingual cultural nuance in machine translation to a large scale review of SOTA model performance across 15 languages. Read the pre-print.

  • Accepted at AAAI 2026 AIGOV & EurIPS 2025 Workshop on Unifying Perspectives on Learning Biases


    This paper explores the vulnerability of several top MLLMs across a range of attack types. Results demonstrate ongoing safety risks in state-of-the-art models and propose the question: is the safest response no response?
    Read the paper

  • Accepted at LSA 2026

    This pilot study explores how well leading multilingual LLMs translate culturally nuanced language—like idioms, humor, tone, and local references—across 20+ languages, using real marketing copy as a test case. Phase two is in progress. Read the paper.

  • Published by The AI Innovator, an online publication focused on advancing understanding and strategic application of artificial intelligence across industries. Read the article.

  • Published by The AI Journal, an online publication exploring the intersection of business and AI. Read the article.

  • Ghostwritten for Appen’s CEO in Forbes. Read the article.

  • This research explores adversarial prompting in LLMs, featuring a benchmarking study of leading models across harm categories. Read the paper.

  • This literature review paper explores a research-based approach to AI safety best practices across the AI lifecycle with examples highlighting AI safety in high-risk industries, such as law and medicine. Read the paper.

  • This work introduces CoT reasoning for LLMs featuring an expert case study on how Appen built a mathematical reasoning dataset for a leading technology company. Read the paper.

  • As a linguistics research intern for Dr. Cory Holland at Colorado State University, my work contributed to her publication on Colorado vowels. Read the paper.

  • I implemented a corpus-based approach to analyzing differences in speech patterns between urban and rural speakers. This work on the corpus analysis tool, with preliminary results, was accepted for a poster presentation at the Western Conference on Linguistics (WECOL) in 2017. 

  • I supported Dr. Sangbok Kim in the development of his Anytime Korean language learning solution as a research assistant and auditor. Learn more.

  • Contributing author to The Language of Localization published by XML Press. Read the book.

Speaking Engagements