CIA Whistleblower: Who Actually Runs The US Government? - John Kiriakou

Chris opens by putting the obvious objection to a former CIA officer on the table: why believe anything he says? Greenberg-style deflections are not available here. Kiriakou's answer is that he cannot compel belief, only point to a record of telling the truth and let the listener decide. He notes the question comes up often and says he will not lose sleep either way — a posture that is either candour or a very practised performance of it, and the interview leaves the judgement with the audience.

Asked about his pursuit of a Trump pardon, Kiriakou answers no, then supplies the structure behind the no. He reports strong support from people inside Trump's inner circle — household names, he says, and he is grateful — but one of them told him it will not happen because the CIA hates him.

His response, three weeks before the recording, was to point out that he has never met John Ratcliffe, the CIA director, and has never exchanged a harsh word with him. The cabinet member's reply is the analytically important part: it is not about Ratcliffe as a person, it is institutional. The fear is that a pardon would open the door for other would-be whistleblowers willing to go public on waste, fraud, abuse and illegality, and the institution is not willing to entertain that.

That answer reframes the pardon from a personal grievance into a precedent problem. The objection is not to Kiriakou but to the signal a pardon would send, which is a different and harder thing to argue against.

Chris asks who is more persona non grata than Kiriakou, and offers Snowden. Kiriakou accepts the ranking and then reaches further back, to Philip Agee, who was still being discussed when Kiriakou was at the agency. Agee was a CIA case officer in South America in the 1960s, in Chile or Peru, who walked in on a torture session run by the CIA together with the host government's intelligence service.

Agee resigned, and then went further than resigning: he wrote Inside the Company, which included an appendix listing the names of every undercover CIA officer he had ever met — roughly twelve hundred names. Kiriakou reports that they wanted to execute him, and that even decades later, when Kiriakou was at the agency, the old timers were still asking how to get Agee. Agee fled to Cuba and lived most of his life there. Until Snowden, he says, Agee was number one.

The exchange then drops into a joke about Andrew Tate and crimes against fashion. It is worth noting only because the tonal shift is characteristic of the whole interview: the material is grave and the register is not.

On presidential control of the CIA, Kiriakou takes a position that is stronger than the usual hedge. The question of how much oversight the president has should not be contested, because the president should have the definitive word. He does not, because there is such a thing as the deep state — and Kiriakou immediately offers the euphemism-tolerant version: call it the federal bureaucracy, call it the state, the name does not matter, the existence does.

The mechanism he describes is tenure asymmetry. Presidents arrive and leave on a four- or eight-year cycle; the career layer stays twenty-five, thirty, thirty-five years. He sat next to a man who is now among the top four or five officers and has been there thirty-seven years. The consequence is procedural: an order they do not want to execute is acknowledged with yes, Mr. President, and then ignored.

Chris supplies the hotelier analogy — permanent staff, transient guests — and Kiriakou accepts it, then adds a second structural problem: the absence of real congressional oversight. The 1975 Church Committee in the Senate and Pike Committee in the House were converted into oversight committees, becoming the Senate Select Committee on Intelligence and the House Permanent Select Committee on Intelligence.

The arithmetic he offers is the argument: how does a staff of forty or fifty oversee an organisation of twenty, thirty, forty thousand? He cannot give the current headcount — it is classified and his information is old — but is confident it is much larger than when he served. The conclusion is flat: you cannot, so you rely on self-reporting.

Kiriakou's illustration is a specific operation that went bad at the very end, in which a life was lost. The rule is that such an event must be reported to the congressional oversight committees. He was driving back to the office with the station chief, who asked whether he was writing the cable, and then told him not to send it to Congress.

The justification is the part that matters, because it is bureaucratic rather than sinister: Congress has enough to worry about, they are working on the budget right now, and we do not want to throw this wrench into the works by telling them we just lost a source. Kiriakou said, you're the boss, and did not report it. He is explicit that this was not unusual.

Chris asks whether that makes the CIA a shadow government, and Kiriakou says yes, that is very much what it is. Asked who actually controls what happens behind the scenes, he locates it precisely: the layer immediately below the political appointees — below the director and deputy director of CIA, the secretary and deputy secretary of state, the secretary and deputy secretary of defense. The career bureaucrats who are not going anywhere.

Chris offers an analogy from running events in the north of England, and it turns out to be doing real analytical work. Nightclub managers wanted the events company's owners standing on the front door next to them — checking the queue moves, that door staff are not assaulting anyone, that the till takings are right — rather than sitting in a car or going for a coffee at one in the morning in the cold.

The point is that presence is not the same as function. The owners were not doing anything operational; the managers knew everyone going in, and Chris was in his mid-twenties while most of the club was nineteen. But what he observed across other events companies with less of a work ethic was reliable: once the owners stopped standing on the door, the events company's managers and the venue's manager began to develop a tighter relationship.

The generalisation is that proximity to information, not seniority, determines who is actually indispensable. The enforcer at the coalface is the one with enough standing to be respected and enough presence to be unavoidable. Chris extends it to the mafia, where bosses have been removed by seconds-in-command and lieutenants, and back to the nightclub: within three to six months, a manager starts calculating that he knows the venue operator and the company is never here, so why not cut the owner out.

Chris then adds the element the analogy lacks: in the intelligence case the owners are replaced periodically, so loyalty cannot even attach to a person. Loyalty runs to an appointee who will be gone in a few years. And the career layer has a vested interest in the status quo because they do not want to lose their jobs — but, Kiriakou adds, more importantly because they are the ones who make the trains run on time. Foreign policy, defence policy and intelligence policy are theirs, and they do not want a bomb-throwing new president arriving.

Kiriakou's sharpest observation in this section is about what the CIA prefers in a president: inexperience. A new president with no foreign policy background is entitled to the President's Daily Brief from the day after the election, and the briefing is delivered by the deputy director for intelligence, the director of the PDB staff, or both.

The staging is the point. The first report laid out is six levels above top secret, everyone else is asked to step outside, and the message is: wait until you see the cool things we are doing around the world. Kiriakou's word for the outcome is recruitment — they made him one of the guys, and thereafter he cannot wait for the briefing six days a week. Someone with foreign policy experience, and perhaps a certain jadedness about analytic or operational failures, is less susceptible to the allure of inside information.

Kiriakou walks the ladder. Unclassified is the New York Times. Sensitive but unclassified is the internal memo — the cafeteria Dunkin' Donuts closing for renovation — sensitive because of where it sits, not what it says. Confidential is most State Department traffic: damaging if exposed only because it would embarrass the federal government. Secret means serious damage to national security; top secret means grave damage. Those are the espionage-law terms.

Above top secret sit the compartments, dozens of them, and the joke is that a new CIA hire does not receive a top secret clearance but something like TS-SCI-SI-TK-GAMMA: top secret, plus special intelligence meaning access to NSA reporting, plus the TALENT and KEYHOLE compartments, with GAMMA above them.

The anecdote he offers for the absurdity is his first classified briefing, where the briefer opened by announcing that the word ZARF standing alone is unclassified. Kiriakou laughed out loud. ZARF was a defunct NSA compartment: you may say the word, you may not say what it denotes. The exchange is a small demonstration of how classification can become ritual rather than protection.

On the operational ceiling, he says GAMMA is the highest on a day-to-day basis — the level at which an analyst can brief Central Command, and the level that would cover something like an intercepted foreign leader's cell phone. Above GAMMA there is a great deal more, and some compartments come and go in months depending on the operation, while others such as ZARF persisted from the 1950s. Code names are computer-generated at random, which is why outsiders cannot infer anything from them.

The compartment discussion ends at DARPA, and Kiriakou uses it to mark the limit of the whole ladder. DARPA's work is at the level of what appears in science fiction films, which he offers as the actual explanation for decades of UFO reports: experimental aircraft, American or the Chinese equivalent. His own limit is stated as an insider's datum — he never met a CIA officer with access to it.

Asked whether the president knows, he gives a two-part answer: if the president asked, he would be read into the compartment and told, but on a daily basis probably not. He then confirms the asymmetry that matters — the president is the only person in government who could obtain access across the board, and the only person who can declassify anything with a stroke of a pen, including GAMMA and everything above it.

Chris asks whether classification is abused to hide embarrassing material, and Kiriakou answers with an unqualified yes. He cites a former classification official under George W. Bush who has since become an anti-classification activist, on the position that fully three quarters of classified documents ought not to be classified at all — the categories being political inconvenience, personal exposure and embarrassment.

The distinction he draws next is the analytically useful one. The White House and the Pentagon leak constantly, but the leaks are almost always authorised: a leak for a political purpose, a leak to embarrass a foreign leader, a leak to test where the public is on an issue. Technically all of it violates the Espionage Act, and it happens every day.

Chris anchors the point in a concrete convergence: a federal judge approving the Paramount–Warner Brothers merger, Larry Ellison's Oracle securing TikTok's US data and algorithm, and Ellison's son David running Paramount including CBS News. He adds the detail that Oracle's name derives from a CIA project its founders worked on. The implication is not a conspiracy so much as a demonstrated permeability.

Asked whether private data and media companies could become as powerful as an intelligence agency, Kiriakou says they already are, then supplies the mechanism that produced it. In the immediate aftermath of 9/11 the CIA sought a congressional waiver allowing it to invest in tech startups — unprecedented for a government, which cannot normally act as an angel investor.

The returns were the point. The CIA was Palantir's very first investor, and Palantir is now a global security software concern with credible reporting of penetration into governments worldwide. Kiriakou's evidence is Sweden, where a national debate over whether to use Palantir for governmental security databases ended with Palantir winning the contract.

His answer to whether Palantir can be trusted not to pass everything to the NSA and CIA is a flat no, supported by fifteen years of service rather than by argument: everything goes to the NSA and CIA. The claim is about architecture, not intent.

Kiriakou's defence of foreign intelligence collection is deliberately narrow. He wants the NSA intercepting Angela Merkel's phone, because he wants to know German trade positions and what the Germans say about the United States. What he objects to is domestic collection — and he notes that both agencies are chartered to be forbidden from collecting on American citizens, and that both do it every day.

Chris raises Flock and the general pattern of using private companies to route around protections meant to restrain government. Kiriakou agrees and dates the change to 9/11. Before it, an FBI investigation required a federal judge and a stated set of reasons; after it, a national security letter to a service provider compels production of everything, and the Patriot Act forbids the provider from disclosing that the letter exists.

The structural consequence he draws is that defence becomes impossible in principle: you cannot contest a search you are not permitted to know about, and there is no judge in the loop to say the line has been crossed. Chris's phrasing is that the safety valve is gone.

Chris puts the harder version of the question: what about private companies building technology that benefits a country, and the relationship that inevitably follows? His example is Palmer Luckey and Anduril, whose tagline he quotes with approval — fight unfair — and whose appeal he states as a preference for a clever person with a gun on your side over a clever person with a gun on theirs.