LLMs exhibit location leakage, generating region-specific content from neutral prompts when location metadata is injected, with rates increasing dramatically even when the location is set to 'Unknown'.
The Geopolitics of Greenland and the Arctic
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3verdicts
UNVERDICTED 3representative citing papers
LLMs exhibit high failure rates on alignment tests for conflict contexts, with some models failing 80-100% on balance requests where legal responsibility is established.
Frontier LLMs in simulated Greenland sovereignty games escalate more under coercion framing, differ by origin when playing the US role, and achieve peaceful US acquisition in only 1.9% of clean games.
citing papers explorer
-
Unintended Effects of Geographic Conditioning in Large Language Models
LLMs exhibit location leakage, generating region-specific content from neutral prompts when location metadata is injected, with rates increasing dramatically even when the location is set to 'Unknown'.
-
Can AI Make Conflicts Worse? An Alignment Failure in LLM Deployment Across Conflict Contexts
LLMs exhibit high failure rates on alignment tests for conflict contexts, with some models failing 80-100% on balance requests where legal responsibility is established.
-
Strategic Coercion Within Alliances: The Greenland Sovereignty Game as an AI Stress Test
Frontier LLMs in simulated Greenland sovereignty games escalate more under coercion framing, differ by origin when playing the US role, and achieve peaceful US acquisition in only 1.9% of clean games.