Geopolitical Divisions Across Languages in Large Language Models
A study finds that large language models exhibit systematically different geopolitical stances depending on the language of the prompt, raising concerns about inconsistent and potentially biased outputs.
A study finds that large language models exhibit systematically different geopolitical stances depending on the language of the prompt, raising concerns about inconsistent and potentially biased outputs.
Auto-scored: The finding is a minor misuse risk, as language-dependent geopolitical bias could mislead users, but it is not a major escalation.