Reasoning of Sovereign Language Models
Event details
| Date | 26.06.2026 |
| Hour | 13:00 › 15:00 |
| Speaker | Ilia Mahrooghi |
| Location | |
| Category | Conferences - Seminars |
EDIC candidacy exam
Exam president: Prof. Robert West
Thesis advisor: Prof. Emmanuel Abbé
Thesis co-advisor: Prof Antoine Bosselut
Co-examiner: Prof. Michael Gastpar
Abstract
The strategic shift toward Sovereign AI addresses the critical need for nations and enterprises to maintain absolute control over their artificial intelligence infrastructure, data privacy, and jurisdictional compliance. As these localized models move beyond basic tasks to complex, multi-step problem solving, two major priorities have emerged: structural safety alignment and the capacity for autonomous self-improvement. By operating in secure, often air-gapped environments, Sovereign Reasoning Language Models (RLMs) must guarantee robust ethical guardrails and scale their intelligence without relying on continuous streams of external, human-curated data.
Selected papers
Exam president: Prof. Robert West
Thesis advisor: Prof. Emmanuel Abbé
Thesis co-advisor: Prof Antoine Bosselut
Co-examiner: Prof. Michael Gastpar
Abstract
The strategic shift toward Sovereign AI addresses the critical need for nations and enterprises to maintain absolute control over their artificial intelligence infrastructure, data privacy, and jurisdictional compliance. As these localized models move beyond basic tasks to complex, multi-step problem solving, two major priorities have emerged: structural safety alignment and the capacity for autonomous self-improvement. By operating in secure, often air-gapped environments, Sovereign Reasoning Language Models (RLMs) must guarantee robust ethical guardrails and scale their intelligence without relying on continuous streams of external, human-curated data.
Selected papers
Practical information
- General public
- Free