Image: Just SecurityAI and the Commercial Data Loophole
• The Pentagon is utilizing commercial Large Language Models (LLMs), creating a "commercial data loophole" that provides the military with powerful surveillance capabilities. • These AI tools can infer sensitive demographic attributes and political ideologies from text, allowing analysts to identify individuals with "unAmerican" views or those likely to organize protests. • This development is concerning because LLM outputs can be inaccurate, potentially leading to severe and unjustified consequences for targeted American citizens.
justsecurity.org