AI safeguards and operational use raise intelligence risk questions

A report says DNI Jay Clayton ordered removal of PRC AI models, including DeepSeek, but implementation is unestablished; PACOM leaders describe AI-agent use in intelligence, reporting, targeting, and warning amid uneven Pentagon safeguards. Claims about an Accenture contractor’s Claude agent potentially exposing credentials remain disputed by Accenture, and compromise is unconfirmed; broader federal AI-inventory research highlights traceability risks without establishing IC-specific effects.