AI Governance Club
Under US Executive Order 14319, a vendor can show that its LLM is ideologically neutral by disclosing the model’s system prompt. That provision assumes the text of an instruction tells you how the system will behave. Anna Neumann, Holli Sargeant, and Jatinder Singh test that assumption.
Join us at the AI Governance Club for session 48, where we will discuss the paper:
Prompt Governance? On Governing Technologies Governed by Natural Language (Neumann, Sargeant, and Singh, 2026)
Session format: Presentation followed by moderated discussion
Presenter: Mariana Hungria
The session is followed by an optional breakout room chat from 18:00 to 18:10 UTC for anyone who wants to stay on and connect.
Mariana Hungria is a linguist and independent researcher working across linguistics, AI safety, and AI governance. She holds a BA/MS in Linguistics, has 20 years of experience in translation and language services, and previously founded and led a tech startup. Much of her recent work has gone into building the field’s foundations in the Portuguese-speaking world. This includes leading the linguistic direction of a 7-million-word translation project covering core AI safety, AI governance, and effective altruism literature. She is Head of Language Services and a field builder at CEGIA, and she co-organizes and moderates the AI Governance Club.
For more information and to register for upcoming sessions, visit the AI Governance Club and follow us on LinkedIn