The CIA is beginning to deploy artificial intelligence to boost one of its most sacred missions: analyzing the plans, intentions and capabilities of foreign countries, the agency’s deputy director said Thursday.<\/p>\n
Since its founding in 1947, the CIA has been the analytical center of the American intelligence community. Its officials analyze fragments of information collected by human spies and covert spying equipment to produce holistic assessments of issues important to American policymakers.<\/p>\n
Now, the agency is starting to rely on AI to make that work harder \u2014 and faster, CIA Deputy Director Michael Ellis said Thursday during an event in Washington hosted by the Special Competitive Studies Project, a nonprofit focused on technology and national security.<\/p>\n
Ellis revealed that the agency recently used AI to create its first autonomous intelligence report, and predicted that the role of AI in its analytical work will only increase.<\/p>\n
Ellis said, “Within the next few years, we will have AI co-workers across all of the agency’s analytical platforms \u2014 a kind of classified version of generative AI that will help our analysts with basic tasks.”<\/p>\n
He said these tasks span the basic elements of intelligence analysis: drafting key decisions, testing findings, and detecting trends in information received by the CIA from foreign countries.<\/p>\n
He predicted that AI could play an even more widespread role in intelligence generation in the future, though he cautioned the agency’s plans to ensure that “humans are the ones making the critical decisions.”<\/p>\n
President Donald Trump and other senior US intelligence officials, including CIA Director John Ratcliffe, have vowed to root out alleged left-leaning within the intelligence community.<\/p>\n
Ellis did not explicitly address that issue. But his speech hinted at how fights over the political legitimacy of intelligence analysis might look different in an AI-centric future.<\/p>\n
At one point, he said the agency “will not let private companies dictate how and when the CIA makes legitimate use of their technologies” \u2014 an indirect jab at AI giant Anthropic, which is locked in a tense legal battle with the Pentagon over its AI guardrails. The Pentagon last month designated Anthropic as a supply chain risk as the company tried to set contractual limits on the use of its models for deadly attacks and mass surveillance.<\/p>\n
The unprecedented maneuver came after Defense Secretary Pete Hegseth called Anthropic “woke” in a social media post, while Trump has accused it of being run by “leftist mad jobs.”<\/p>\n
America’s premier spy agency is also adopting AI for many other tasks beyond analysis.<\/p>\n
The CIA tested 300 AI projects last year “to bring new capabilities to our mission,” such as processing large datasets and language translation, Ellis said. He said it is also trying to get AI and other cutting-edge technology into the hands of its foreign officials who collect top-secret information on the military, political and economic activities of foreign powers.<\/p>\n