Emphasizing a strategic pivot toward cyberoperations, C.I.A. Director John Ratcliffe assured lawmakers Tuesday that the agency’s new machine-learning tools will remain under the strict control of human personnel uniquely trained in unconstitutional regime change.
WASHINGTON — Testifying before the Senate Intelligence Committee on the agency’s modernization efforts, C.I.A. Director John Ratcliffe stated Tuesday that the intelligence community is prepared to take “smart risks” in the digital domain, provided that its deep-learning models are never allowed to act unilaterally without a human operative in the loop.
According to unclassified transcripts of the hearing, the director insisted that any automated proposals to fund paramilitary death squads, traffic illicit arms, or surveil domestic journalists would first be routed through a veteran case officer who could properly file the classified paperwork. The sweeping reorganization is designed to streamline the Directorate of Digital Innovation by allowing algorithms to process vast amounts of illegally intercepted communications, leaving human agents free to focus on the traditional tradecraft of physically bribing local officials and categorically denying American involvement to the global press.
While our neural networks can optimize a Latin American destabilization campaign in microseconds, it is essential that an American citizen is sitting at the desk to officially perjure themselves before Congress when the time comes.
Following the briefing, sources close to the committee indicated that lawmakers were largely satisfied by the C.I.A.'s commitment to maintaining a human element in its black-sites program. Civil liberties advocates similarly expressed cautious optimism regarding the new operational guidelines, noting their profound relief that the unfettered power of artificial intelligence would be kept firmly in check by the rigorous ethical framework of the institution responsible for MKUltra.