While missiles hit Iran this week, Oracle quietly announced that the US government authorized it to run generative AI on federal data. Not some sandbox test. FedRAMP High with DISA Impact Level 5 — that’s Department of Defense mission data, classified intelligence systems, and Medicare records for over 150 million Americans. The AI model running across all of it? Grok, from Elon Musk’s xAI.
In the same ten-day window: Oracle’s consortium closed the Paramount-Warner Brothers Discovery merger — $111 billion, the largest media deal in American history. Oracle won an $88 million Air Force cloud contract. And it picked up the CMS contract for Medicare and Medicaid data.
One company now sits beneath American defense infrastructure, intelligence systems, healthcare records, and the largest media company ever assembled. And the person Oracle hired to lead its health division is Seema Verma, who previously ran CMS itself — and was investigated by the HHS Inspector General for mishandling millions in contracts during her tenure.
This isn’t new behavior. Oracle’s first customer was the CIA. Between 2014 and 2024, it assembled one of the most powerful data collection operations in the world through a series of acquisitions — BlueKai (invisible tracking pixels), Datalogix (offline purchase data), and AddThis (share buttons on 15 million websites reaching 1.9 billion users monthly). That operation built “identity graphs” — profiles connecting real people to their behavior across every device, every website, every purchase. Larry Ellison told his own investors in 2016 that Oracle had data on over 5 billion people. A federal lawsuit and $115 million settlement later, they shut down the advertising side. But the surveillance capability didn’t disappear. It just found a better customer: governments.
This is exactly the scenario that Universal Basic Citizenship is designed to prevent. Not through protest, but through infrastructure. When we talk about democratic infrastructure, we mean the systems that keep power distributed and accountable. Public AI that serves citizens rather than surveilling them. Data sovereignty that treats your medical records and your browsing habits as yours. Governance structures where no single company can sit beneath defense, healthcare, media, and intelligence simultaneously without meaningful oversight.
The Dre Dossier, which broke this story together with detailed sourcing, has written a 45-citation policy brief asking Congress for three specific things: a formal conflict-of-interest review of Oracle’s overlapping government contracts, data firewall requirements between Oracle’s federal and commercial operations, and reclassification of Oracle as a related party rather than a neutral vendor. That brief has been sent to every member of the Senate Antitrust Subcommittee.
This is the kind of concrete policy work that matters. Not just naming the problem, but proposing structural solutions. Foundation’s position is that this consolidation is exactly what you get when you don’t have democratic infrastructure — when there’s no public AI alternative, no data sovereignty framework, no mechanism for citizens to have a say in who controls the systems that run their lives.
The window for getting this right is narrowing. Every week that passes with AI authorization expanding and oversight shrinking makes the next week harder. That’s not alarmism. That’s architecture.
Source and further reading: The Dre Dossier — “The Merger That Needed A War” and the full policy brief at thedreydossier.substack.com.