Administration officials have asked a government AI testing unit to halt public reports, sparking concerns over national security and control.
The Center for AI Standards and Innovation (CAISI) was instructed to stop issuing public assessments while an executive order is implemented.
CAISI's work is seen as crucial in mitigating AI risks, including cybersecurity threats and biological weapons.
The move has thrown the unit's future in jeopardy and raised concerns among companies like OpenAI and Anthropic.
AI analysts warn that an overzealous model-testing process could slow deployment and hinder innovation.
The White House has defended the decision, citing the need to protect national security and promote innovation.
