CIA's Newest Interns Don't Need Sleep: Agency Welcomes AI Co-Workers to the Intelligence Grind

The US Central Intelligence Agency is getting ready to onboard some very dedicated team members—ones who never complain about Monday mornings, ask for raises, or demand better parking spots. Deputy director Michael Ellis announced at a Special Competitive Studies Project event in Washington that the agency plans to embed AI co-workers directly into its analytics platforms within the next couple of years. These digital interns won't unionize, won't take sick days, and apparently won't judge analysts for eating cold pizza at their desks at 2 AM.

These classified generative AI tools will help analysts with the grunt work: drafting key judgments, stress-testing analytical conclusions, and spotting trends in the mountain of intelligence the agency collects from abroad. Ellis was quick to clarify that humans will still be the ones making the actual calls—because apparently, Skynet jokes write themselves. The machines get to do the boring stuff while humans pretend they still matter. Efficiency!

The timing is interesting. The CIA's AI push comes amid a heated feud between the Department of Defense and AI firm Anthropic. Despite a $200 million contract, Anthropic drew the line at letting its Claude model power mass domestic surveillance or fully autonomous weapons. President Trump ordered all federal agencies to stop using Anthropic's tech in March, and the DoD labeled the company a supply chain risk. The legal brawl continues, with a US appeals court recently nixing Anthropic's request to pause that designation. Nothing says "we're definitely not retaliating" quite like a federal appeals court conveniently siding with the Pentagon.

Ellis didn't name-drop Anthropic, but made clear the agency won't let any single company call the shots: "We cannot allow the whims of a single company to constrain our capabilities." Translation: we're not getting ghosted by another AI startup that suddenly develops ethics. The CIA has been burned before—well, not burned exactly, but definitely left on read.

The CIA has already been dabbling. Last year, they tested roughly 300 AI projects for processing large datasets and language translation. They even produced their first fully AI-generated intelligence report—a milestone that probably made some analysts nervously glance at their job descriptions. Nothing like knowing your performance review might soon be written by the same tool that's helping you write it. Promotions all around!

Staying ahead of China is a major driver. Ellis noted the technological gap that once existed between the US and China has basically evaporated. "Five to ten years ago, China was nowhere near America in terms of technological innovation. That's just not true today." The vibes have shifted, the spreadsheet has updated, and everyone's pretending they saw this coming.

In case anyone forgot this was a crypto newsletter, Ellis also pointed out in May that Bitcoin and crypto are national security matters. The agency uses blockchain data for counterintelligence operations—because apparently, on-chain analytics isn't just for degens tracking whale wallets anymore. Somewhere, a chain analyst is finally getting respect at dinner parties. "No, mom, my blockchain tracking skills ARE nationally relevant now."