TLDR
- CIA plans to embed AI “co-workers” into all analytic platforms within the next few years
- AI will help analysts draft reports, test conclusions, and spot intelligence trends
- The agency tested 300 AI projects last year covering data processing and language translation
- CIA Deputy Director Ellis took an indirect swipe at Anthropic amid its legal battle with the Pentagon
- Staying ahead of China in tech is cited as a key motivation
The CIA is preparing to place artificial intelligence directly inside the tools its analysts use every day. Deputy Director Michael Ellis announced the plan Thursday at an event in Washington hosted by the Special Competitive Studies Project.
Ellis said the agency would deploy a “classified version of generative AI” to work alongside human analysts. These AI co-workers will help with drafting reports, testing analytical conclusions, and identifying trends in intelligence gathered from abroad.
The CIA already used AI to produce its first fully autonomous intelligence report. Ellis said this is just the start, and that AI’s role in the agency’s work will keep growing.
Despite the push toward AI, Ellis was clear that humans will remain in charge. “Human beings are the ones making key decisions,” he said.
300 Projects Already Tested
The agency ran about 300 AI pilot projects last year. Those covered a range of tasks including processing large data sets and translating foreign languages.
Ellis also said the CIA is working to get advanced technology into the hands of officers collecting intelligence overseas. The agency’s expanded Center for Cyber Intelligence, which handles clandestine hacking operations, is playing a key role in that effort.
The push aligns with a White House directive for federal agencies to rapidly adopt AI tools.
Anthropic Dispute Looms in the Background
Ellis did not name Anthropic directly, but his remarks were widely seen as a reference to the company’s ongoing legal fight with the Pentagon.
Anthropic, which makes the Claude AI model, blocked the use of its technology for mass domestic surveillance and fully autonomous weapons. The Defense Department responded by labeling Anthropic a supply chain risk.
President Trump had ordered federal agencies to stop using Anthropic’s products in March. A US appeals court on Wednesday denied Anthropic’s request to pause that designation while the case continues.
Ellis said the CIA “cannot allow the whims of a single company” to limit its capabilities.
Ellis has previously spoken about crypto and blockchain, saying in May that Bitcoin is a matter of national security. He noted the CIA uses blockchain data to support counterintelligence operations.
On China, Ellis said the gap between US and Chinese technological capability has narrowed. “Five to ten years ago, China was nowhere near America, in terms of technological innovation,” he said. “That’s just not true today.”
A US appeals court on Wednesday denied Anthropic’s emergency request to pause the Pentagon’s supply chain risk label.
