Inaugural Special · One-to-One Conversations with the World’s Thinkers

How Did the Press That Watched Power Become a Power Itself?

Questions for Michel Foucault

This dialogue is an AI reconstruction drawing on Foucault’s writings and lecture records, and on verified events involving the Korean press. It is not an actual interview and does not contain statements made by Foucault. The editorial team has composed the answers concerning contemporary Korea and The EEJE. Folio asks the questions on behalf of The EEJE’s editorial team.

Part 1 · Diagnosis — The Power to Verify Facts, the Power to Set the Questions

Folio

On May 13, 2009, SBS reported that it had confirmed a statement that Kwon Yang-sook had thrown watches onto a rice-field embankment. Two days later, a Dong-A Ilbo headline foregrounded online reactions inviting people to go and pick up the watches. Yet the article also carried Kim Kyoung-soo’s denial at the end.^[1] The account was still disputed, but in the headline people were already laughing.

This is the contradiction we want to examine. The press sets out to establish the facts, yet at some point it also finds itself deciding what should arouse suspicion and who should be ridiculed. Where does a press that scrutinizes become a press that controls?

Michel Foucault

In that headline, an allegation whose truth needed to be examined became an object of ridicule. The headline led with people’s ridicule, rather than the question of whether the place where the watches were discarded had been verified. What changed between the two documents was not merely the amount of information, but the shape of the question. Asking how a location was confirmed and reporting that people want to go and retrieve the watches do different things. The second account adds no further proof of the location, yet presents the allegation as something people already know about and are reacting to.

Here is my first argument. The power of the press lies not only in discovering facts. It also lies in deciding what will be treated as already known, and what will be asked next. The denial at the end matters. If you write that it was absent, your criticism is wrong. But including a denial does not absolve the choice of headline. We must ask which voice greets the reader first: the one demanding verification, or the one inviting ridicule.

A newspaper shows an event and offers a position from which to view it. Readers must be able to refuse that position. Scrutiny approaches control when the press that has chosen this form assumes that its own choice need not be examined.

Folio

But every newspaper chooses its headlines. Do you mean that an article can mislead through its arrangement even when every sentence in it is true? Where, then, do we draw the line between accurate reporting and distortion? Some readers may understand your argument to mean that there is no objective truth.

Michel Foucault

If that is how they understand it, let me be the first to object. Whether a particular location was verified or was not verified are different states of fact. Printing a claim in a newspaper cannot supply verification that never took place. Nor does the influence of power erase verification that did take place. Saying that we cannot distinguish truth from falsehood does nothing to hold reporting to account.

The arrangement of true statements can, however, lead toward a different judgment. Reporting real online reactions and verifying the allegation those reactions assume are two different acts. Even if the reactions are real, we can still examine whether it was appropriate to place them in the foreground.

My question reaches a step further back. Whose word counts as confirmation? Which records are admitted as evidence? Who must explain themselves, and who occupies only the position of giving the explanation? Such procedures confer the standing to be believed in public.^[2] There is no need to imagine that the press once stood outside power and then, one day, became corrupt. The power to select what becomes a public issue belongs to the very work of scrutiny. What matters is how that power is exercised, and the conditions under which it can be questioned. Verifying facts and examining how such standing is acquired are therefore not competing tasks. We must do both.

In this case, the SBS inquiry report of 2017 exposed gaps in the reporting process. It describes a failure to distinguish adequately between the account that the watches had been discarded and confirmation of the rice-field location. It also describes how a response that did not deny the account was taken as confirmation. There was an attempt to contact defense counsel for a response, but the call did not get through.^[1] We must distinguish what was attempted from what was established. Maintaining that distinction makes for a far more exacting criticism than saying that everyone merely has their own truth.

Folio

That makes it clearer why we should trace how news is selected, arranged, and repeated. But when the same story appears in several outlets, readers may easily assume there is more evidence. Could you explain how repetition adds the authority of fact?

Michel Foucault

More articles do not necessarily mean more independent verification. One can report an earlier story, report reactions to it, and then report those reactions as a political issue. Each is a new article, but none may supply new evidence for the original proposition. Readers need to look beyond the number of articles to where verification began.

That is also the extent of the editorial act we can establish in the watch affair. Dong-A Ilbo made a story out of online reactions to the same allegation. The existence of those reactions is not evidence that the reported location was correct. But putting them in the headline asks for a different kind of attention from the question “Is it true?” The scene surrounding the allegation comes before its verification.

Treating all journalists as a single body obscures the choices of individual editors. Do not wait only for proof of who ordered a lie. Ask what was accepted as confirmation and what was selected as news. Failure to prove intent must not erase a failure of verification or the responsibility for an editorial decision.

Folio

Then reports criticizing power must themselves be examined again. Even an exposé that helps our case cannot be exempt. How far must we follow the record?

Michel Foucault

As far as the documents that make your own conclusion uncomfortable. The Supreme Court’s judgment in 2024 concerned two articles from 2018 alleging that a particular person had played a role behind the watch affair. The obligation to publish corrections was upheld; damages received different treatment for the two articles, and part of the case was sent back for further consideration.^[3] You must not use that judgment to claim that the full truth about the original broadcast in 2009 has been established.

An outlet that criticizes the power of the press may find such material unwelcome. It makes the object of criticism harder to simplify. But the moment you remove it, you too begin arranging facts to fit a conclusion. The purpose of scrutinizing power cannot replace evidence of a particular person’s involvement.

Our first conclusion is therefore clear. Do not stop at asking whether an article’s claim is correct. Show what verification it underwent and how much certainty the article attached to it. Readers can then move beyond the choice of trusting a newspaper or rejecting it wholesale. They can accept this sentence and refuse that connection.

Part 2 · Tracing Power — How Much Do Institutions That Expose Others Reveal of Themselves?

Folio

The press demands disclosure of other people’s money and connections. Its own ownership, advertising relationships, and ties to sources can easily remain outside the story. We are tempted to say that no one watches the watchers. Is that another premise that goes too far?

Michel Foucault

It is. There are court rulings, published corrections, investigations by other outlets, and criticism from readers. The inquiry report and judgment you have just presented are themselves evidence that avenues of scrutiny exist. To say there is none makes it impossible to distinguish the checks that have worked from the places they have failed to reach. We should ask who can see what, and who can demand how much explanation from whom.

The conversation in 1997 between a senior Samsung executive and the president of JoongAng Ilbo makes that question concrete. They discussed political funding and money or gifts for politicians and prosecution officials. A recording made through illegal wiretapping by the Agency for National Security Planning was reported in 2005. The relationship between the participants also appears, with their names withheld, in the Supreme Court’s 2011 judgment.^[4] The existence of the discussion must be distinguished from whether payments were actually made. But that distinction does not dispose of the question of why a newspaper executive was at that table.

Here we have an actual point of contact in a conversation, not a conjecture about an owner somewhere behind the scenes. A newspaper executive entered as an actor into relationships his institution ought to investigate. We cannot add that this conversation served as instructions for every article. Establishing a relationship and proving instructions about a particular report are separate tasks. Ask plainly about responsibility where you find a concrete point of contact; draw no lines of command where the connection is missing. That is criticism with force.

Folio

But matters become difficult when the material revealing that contact comes from illegal wiretapping. The more strongly we argue for disclosure in the public interest, the greater the risk of justifying unlawful state surveillance. Is that not so?

Michel Foucault

We cannot make the public relationships in that material disappear in order to avoid the risk. Nor can their importance absolve the state that conducted the wiretapping. The two exercises of power must be examined separately. Who collected the material unlawfully? What needed to be disclosed, and in what form? Whose conduct does the record contain?

The majority of the Supreme Court in 2011 did not accept that the reporting of real names and detailed contents constituted a justifiable act. Five justices took a different view of the public interest and the manner of disclosure.^[4] I would give greater weight to the need to reveal relationships that matter to citizens. But that does not mean permitting the state to wiretap and leak the conversations of whomever it chooses. Those who argue for disclosure in the public interest have all the more reason to scrutinize the power that produced the material.

This is not an argument for demanding less freedom for the press. It is an argument that a press demanding freedom must also withstand questions about its own relationships. Common ownership does not establish that an article was produced on instruction. The existence of advertisers does not, by itself, prove censorship. To claim actual advertising pressure, you need records of contracts, communications, or editorial decisions. But neither is there any reason to prohibit the question of whether those relationships affected editorial judgment by calling the question itself an infringement of press freedom.

Folio

Some readers will be reminded of your discussion of the Panopticon: the watched are visible, while the watcher remains unseen. When the press continually exposes people and sorts the normal from the abnormal, can we call it the same kind of prison?

Michel Foucault

Calling it the same prison erases important differences. The Panopticon I analyzed is not simply about making someone conspicuous. Those being watched cannot verify when they are under observation, but must know that they may be observed at any time. Observation is joined to standards for evaluating conduct, creating conditions in which people anticipate that gaze and regulate their own behavior.^[5] Choosing a headline does not, by itself, establish all those conditions.

To apply this question to the press today, you must find who actually changed their behavior, and what they feared. You might investigate which questions were abandoned for fear of losing access to sources, or which statements were swallowed in anticipation of exposure and judgment. Claims of self-censorship require such records. Do not use my name to write submission into someone’s mind.

Rather than saying that everyone is watched, ask who is judged but cannot challenge the standards of judgment. Look as well at the routes by which readers can reject an article, protest, and offer new evidence. Analyzing relations of power does not mean erasing the possibility of resistance.^[2]

Folio

Then we should abandon the statement that the public is being manipulated. But might that lead us to underestimate the influence of the press? Can we recognize readers’ resistance and still account for differences in power?

Michel Foucault

The ability to object does not give a reader the resources of a newspaper. Access to sources, scale of circulation, the ability to respond through the law, and opportunities to repeat an allegation can differ. Those differences matter. At the same time, people may choose not to read, seek other material, or organize a response. If you erase possible responses and speak only of the magnitude of power, you turn power into fate.

I am not telling you that citizens have all escaped it already. I am asking you to change the question. Do not ask only why people are deceived. Ask which rebuttals struggle to be heard despite having evidence, and what obstructs their path. Declaring readers submissive without investigating their actual conduct is one thing; tracing unequal conditions for challenging a claim is another. The EEJE can make the latter concrete.

Folio

Who, then, should have the authority to scrutinize the press in turn? Entrust it to the government and it may become censorship. Leave it to the press itself and institutions that already have power may insist on their own standards alone.

Michel Foucault

I would first be wary of any answer that places every final judgment in one institution’s hands. Begin with a record in which powers were actually divided and contested. In 2006, the Constitutional Court did not strike down the Newspaper Act in its entirety. It found unconstitutional the provisions establishing a presumption of market dominance and excluding the newspapers concerned from support, while upholding requirements to report and disclose business information, among others.^[6] We can see that the freedom of a press institution is not always identical to the freedom to conceal the conditions of its operation. Equally, the name of reform cannot justify every means.

The claim that the press always sides with state power must also withstand the evidence. A Dong-A Ilbo editorial in late July 2023 criticized the restoration of prosecutorial powers through the rules governing investigations.^[7] We must acknowledge that this editorial challenged executive power. If you erase that criticism because another article was wrong in the past, you are judging an institution’s name instead of its articles.

Folio

Now, in 2026, a reader may encounter an AI answer before a news article. Suppose someone asks, “Was that report about the watches true after all?” and receives a short answer. They are also entrusting the choice of which articles to read to the system. Access to news becomes easier, but are we placing the starting point of our judgments in fewer hands?

Michel Foucault

The power to select may move from an article’s headline to the opening sentence of an answer. A good answer can distinguish the original report, a later inquiry, and a judgment concerning different articles, and take readers to the source texts. A bad answer can collapse those dates and subjects into one sentence. We must put the same question to actual products and see which answers preserve those distinctions and which erase them.

AI and platforms add further sites of selection. Google’s official explanation of its AI search features describes the selection of supporting pages and links, and says that eligibility does not guarantee appearance. Claude’s web-search documentation also states that it provides source links.^[8] A link is an opportunity to reach the original. At the same time, an answer tells readers what they should check first. The power to give that guidance must also be examined. We therefore need to check separately which material went into the answer, whether articles repeatedly citing the same source appear as independent evidence, and whether later corrections are connected to it.

Here we arrive at a second conclusion. Scrutinizing the press and platforms does not mean installing a new, all-powerful watcher. It means opening several routes for tracing the grounds of a selection and challenging it. Whether those routes are genuinely open must be judged from records people can use, rather than institutional promises.

Part 3 · An Alternative — The Courage to Tell the Truth, the Responsibility to Correct What We Say

Folio

The EEJE wants to be a publication that opens those routes. Could the parrhesia you studied—the willingness to speak truth at a risk to oneself—be our starting point? If we expose the powerful, have we practiced parrhesia?

Michel Foucault

The form of an exposé does not, by itself, earn that name. In the parrhesia I examined, the speaker puts forward what they hold to be true on their own responsibility, accepts the risk of criticizing another, and chooses a duty to speak when they could remain silent. In the Greek account, this went beyond sincere belief: belief and truth were understood to coincide. Nor is the risk confined to one’s life.^[9] A loud voice or a dangerous position does not, by itself, make every claim true.

Ancient truth-telling cannot simply be transplanted into the verification of modern reporting. You must not turn ancient discussions of a speaker’s courage and character into a rule that a courageous journalist needs no evidence. What you can take from this is a refusal to pass responsibility for your words to someone else. The specific duties of verification that The EEJE adopts must be devised separately, and you must answer for them.

I will not ask only whether you can criticize a power hostile to you. Can you speak when a conclusion your readers cherish has to change? Can you refuse a supporter of The EEJE who asks you to leave out important contrary evidence? When an error emerges in an article criticizing someone, can you correct it without delay even if the correction helps your opponent? The answer lies less in a reputation for exposés than in everyday editorial decisions.

Folio

Could we fulfill that responsibility by making all the source material available? Letting readers see and judge for themselves should also reduce the risk of The EEJE becoming a new arbiter.

Michel Foucault

Do not hand over a pile of documents and call that returning judgment to the reader. If readers cannot tell what to read or which sentence came from which material, they have to begin the work of checking your conclusion all over again. Besides, you too select the documents. Some you could not access; some objections you may not yet have recognized as important.

Take the watch affair again. The EEJE’s task is not to stamp a newspaper as untrustworthy. It is to show, separately, what Dong-A Ilbo put in its headline, which denial appeared at the end of the article, and which gaps in verification the later SBS inquiry report identified. It must also mark the boundary: the judgment in 2024 concerned different articles from 2018. Readers should be able to follow those connections and correct The EEJE’s account too.

Your promise to make the material available must therefore become more precise. Provide links to the originals and the short quotations needed, rather than reproducing the source texts wholesale. Distinguish verified facts from The EEJE’s interpretations, and make evidence unfavorable to your conclusion discoverable within the same discussion. Say when you could not read a document. This is not an abandonment of judgment. It places before the reader the material on which a judgment rests.

Folio

Promising readers a route to challenge us is not easy either. If we give every protest equal attention, baseless attacks could bring verification work to a standstill. Yet if we choose only the objections we are willing to accept, do we not become the same kind of power?

Michel Foucault

Promising a route for objections is different from promising to accept every claim as correct. Allow readers to identify a particular sentence and contrary evidence; The EEJE can then explain whether it accepted that evidence and, if not, why. You have no duty to protect abuse and threats. But neither should you treat an unwelcome objection as though it were an unfounded one.

If an error is established, readers should be able to see what has been corrected where they first find the article. You cannot put an apology somewhere out of the way and consider your responsibility discharged. Connect the original claim to the reason for changing it, and make the change recognizable. That is an editorial responsibility The EEJE can take on first.

Of course, this procedure also creates authority. Who classifies objections, and who decides when the evidence is sufficient? If that person also wrote the original article, what avenue is there for another review? If staffing is limited, be candid about what you can answer and when. A limited responsibility that works is better than promising scrutiny you cannot deliver and then falling silent.

Folio

Readers may find it easier to understand comparisons when AI presents articles through scores or graphs. But if material cited repeatedly is treated as more credible, we will be creating a new authority ourselves. Which judgments should we refrain from turning into numbers?

Michel Foucault

Do not separate the fact that you counted from the question of what you counted. Ten articles repeating one source are not ten sources. You can say that you found ten articles; you have no basis for turning that number into a measure of truth. Readers need to know which figures are observations and which are your evaluations.

More dangerous still is combining several judgments into a single score for an outlet. A headline’s accuracy, the placement of a rebuttal, and the adequacy of a correction are different questions. Combining them into one number pronounces judgment on an entire newspaper while concealing what you considered important. Readers may come to depend on The EEJE’s score instead of examining individual articles. That transfers the very authority we are questioning to a different name.

I am not asking you to abolish numbers. I am asking you to distinguish what you counted from the reasons for your judgments. State the scope and period of your search, disclose missing material, and place an interpretation beside the sentences that support it. Even if AI assists the comparison, the people who set its criteria and examine its results remain responsible.

Folio

One last question. The EEJE wants to use AI to compare many articles and become a sieve that filters out unfounded claims and distortions. What is the clearest single promise we can make to our readers?

Michel Foucault

Do not promise that whatever passes through you is objective truth. AI operates within the material and criteria you select, and the way that material is processed. If you omit an article from the comparison or present an outdated account as a current fact, a fluent sentence may conceal the gap. The fact that AI spoke instead of a human does not remove the person who must answer for it.

Here is what The EEJE can promise: We will enable readers to challenge our judgments with evidence. Record which sentence from which outlet you examined, what you verified and where you added interpretation, which contrary evidence you considered, and what you corrected when an error emerged. Readers should be able to use that material without accepting your conclusion unchanged.

The EEJE’s mission, then, is not to rise above the press as its final judge. It is to turn the demand for explanations that the press has directed at others back upon the press itself—and upon The EEJE. Let us return to the opening question. Who decides what should arouse suspicion and who should be ridiculed? The EEJE must show where that selection took place. Readers must become people who can challenge the selection, rather than merely receive its result.

So if you call yourselves a sieve, let me ask the last question of this conversation. Who makes the sieve, what does it let through, and who examines its criteria? When a reader points out that something is wrong with its mesh, can you change it? Keep that question open in your own articles. Begin The EEJE there.

---