Disclaimer: The views and opinions expressed belong solely to the authors and do not necessarily reflect the official policy or position of any previous employer or organization with which the authors have been affiliated.
Background: The Information Wars is a conflict between those on offense, seeking to covertly influence public opinion for ideology or profit, and those on defense, seeking to make public these secret campaigns and defend verifiable, accurate information. Those on offense seeking profit, companies and individuals that act (mostly) legally, can make billions. The fraudsters are left to law enforcement. What remains are political actors, both state and non-state, the tracking of which becomes the responsibility of academia, NGOs, and governments.
Context: The following is a summary of conclusions from those tasked with defense in the Information Wars, tracking and uncovering foreign-origin state and non-state influence operations targeting elections, public health, and other core aspects of society deemed crucial to national security.
These perspectives originate from work done for the U.S. Department of State. The U.S. government is vast, other departments with different authorities may have had different rules of engagement and conclusions. However, we were the primary group working on foreign influence operations against the U.S. and its allies. Our analysis and conclusions, while not exhaustive, carry the most firsthand experience.
This work is meant solely as a retrospective analysis, references to rules of engagement, cultural norms within government work, etc. may have changed as of 2025.
---
Lesson 1: If the Message is Right, the Messenger Does Not Matter
In tracking foreign influence operations, several assumptions were made. We assumed the public would want to be aware of fakery and fraud. We assumed people cared if they were being manipulated. We assumed American politicians would unite in opposition to foreign attempts to covertly influence American public opinion. We assumed most people cared if the content they enthusiastically disseminated to their own trusted social graph was factually accurate, or if the person behind the account was real.
We were wrong. And in retrospect, naive.
The phrase “politics makes strange bedfellows” was an old way of describing how, in politics, the enemy of my enemy becomes my friend. In 2026, this phrase does not meet the moment anymore.
In the Information Wars, if “Country X”, a nation with a clearly articulated history of degrading and destroying your country, happens to support your preferred candidate in an upcoming election, they become an ally. Partisans enthusiastically promote and disseminate Country X’s state media, embracing its bots and trolls because they, in this narrow sense, have the same goal you do: get your preferred candidate elected.
This dynamic hints at a greater compromise: a Faustian bargain with your country’s historical (and contemporary) enemies for short-term political gain.
The reality is, there is no short-term gain working with a country whose main goal is the long-term defeat of your country. Will those who made this bargain ever have a moment of introspection as to why their country’s unambiguous enemies support their preferred candidate? If you truly want what is best for your country, how can your desired outcome align perfectly with somebody who indisputably does not?
Perhaps one day, but we are not holding our breath.
This first draft of this retrospective avoided this point entirely. Government work requires nonpartisanship. Producing work that appears opinionated or not based solely on factual reporting is considered poor quality.
Practically, politics should be avoided in order to speak to the largest possible audience. In analysis, any specific politicized content would put at risk the integrity of the overall message. However, you cannot discuss the Information Wars without discussing how information has been weaponized by partisans in the pursuit of power, unconcerned with the outcomes and consequences of embracing conscious and purposeful deception.
Lesson 2: The Elephant in the Room: The Information Wars Are Not Apolitical, They Are Inherently Political
The public remains largely unaware of an important cultural staple of government work: all government personnel have it drilled into them from the beginning to be strictly apolitical. One of the worst things to be accused of, either in diplomacy or national security, is to be biased in your work. Even the appearance of bias could compromise your work and your career. With a new administration, this attitude may no longer be accurate, but just a few years ago, it was the reddest of red lines.
Understanding the politicization of truth remains core to the understanding of how the Information Wars have unfolded, so it must be confronted.
Yes, there is evidence that hostile foreign influence operations supported candidates from both the left and the right, but crucially, only candidates foreign nations viewed as unviable, divisive, and ultimately destructive. Bernie Sanders is the example from the left, and Donald Trump is the example from the right. In 2016, Bernie Sanders’ candidacy received foreign support which increased only after he had lost his primary to Hillary Clinton, with the goal of promoting resentment towards the democratic process. Ultimately, those foreign resources coalesced around the candidacy of Donald Trump.
In our work, we imagined a cat-and-mouse game between governments. But in reality, when we exposed foreign governments, they deleted their tracks and started over. The biggest adversary we faced was often our own government, looking to obstruct, degrade, and eventually destroy our ability to track foreign influence operations against the United States.
During COVID, it was administration policy to slow down testing in order to have fewer cases. This is, of course, absurd because if you do not test, you are just unaware of what is happening. Obviously, this is not the same thing as reducing infection. The administration took this same approach towards foreign influence operations. The less we know, the better.
Unfortunately, the dynamic we saw firsthand cannot be explained only by the desire for plausible deniability. Our jobs tracking hostile foreign influence operations were made nearly impossible by coordinated attempts by our own government to slow down and stop our work entirely. Endless FOIA requests, congressional inquiries, and baseless accusations from either misinformed or malicious congressmen. All from one party.
Right-wing members of the U.S. Congress actively degraded our ability to track foreign influence operations because they saw these operations as an asset to their party’s de facto leader, not the behavior of an enemy.
Fast forward to 2025, the current administration has dismantled every domestic capacity to track foreign influence operations. Obstruction had turned into elimination.
Those on defense were wiped out entirely, evoking the rebels at the end of Star Wars: The Empire Strikes Back. The battle was lost, the resistance scattered, and no one knows what will come next.
An unceremonious end.
We lost the Information War, not because of a decisive victory by the enemy, but because of extended sabotage from within. The defeat was not just to us, or our profession, but to our country. Nations on offense increase their budgets for information warfare every year, and we voluntarily surrendered our defense. Programs set up to work with allies were dismantled, even physically banishing the individuals involved. This was not just a change in policy, this was purposeful destruction to ensure defeat.
Lesson 3: Defense in the Information Wars Requires Good Faith, Offense Relies on Bad Faith
Plainly stated, defense in the Information Wars depends on people caring about truth. If a critical mass of the general population becomes so cynical that truth no longer matters, playing defense becomes pointless. This is the end state of a society that does not prioritize truth, to be governed by amoral leaders who view the truth as, at best, inconvenient, and at worst, an impediment to power.
The Information Wars are a battle for the survival of democracy, which is why so many authoritarian countries choose to play offense against democracies, or why anti-democratic forces within a democracy embrace lies so openly and brazenly.
Is it their way of ensuring job security.
The more toxic and bad faith online discourse becomes, the harder it is to play defense. However, it is misleading to blame outcomes of mistrust and rising extremism solely on foreign influence operations. There is a large, domestic, fully organic aspect to the problem.
Lesson 4: The Problem is Not (Just) Supply of Bad Information, the Problem is (Also) Demand
We are going to say what the owner of every social media platform wishes they could admit: often, the people are the problem.
Playing defense by exposing attempts at manipulation and distortion becomes pointless if dishonesty is tolerated, even preferred, as long as the message is correct.
A narrative by itself does nothing if it does not speak to something in the audience. You cannot simply put out a message you want others to believe, it will be drowned out in the sea of endless digital noise.
In order to execute a successful influence operation, you need to say the right thing to the right person, at the right time. You need to speak to people, in their language, from the perspective of their worldview. You need a deep cultural understanding of what is motivating a population in order to exploit it. The right message also depends on perfect timing, an accurate message delivered too early or too late does not resonate, nor go viral.
Those on offense know their job is to find where the interests of their government overlap with the interests of a target population. Once you have found that intersection, amplifying shared narratives will find willing and able collaborators within your target audience. In other words, your target will do the work of spreading your narratives for you.
This is the sweet spot for foreign influence operations. Persuasion goes from difficult to frictionless, not out of coincidence or luck, but out of genuine common interest. A successful influence operation does not “trick” anyone. Instead, it succeeds in speaking to the precise part of a target’s ideology that aligns with your own. In this way, successful influence operations do not resemble one side speaking to another, but an active conversation.
A mutually beneficial partnership.
Confirmation feels good, rejection feels bad. Therefore, it should surprise no one that foreign-based fake accounts, focused solely on partisan messaging, amass huge followings. It is no wonder people would rather listen to foreign state media in support of their ideology than domestic media that challenges it.
It is human nature.
Lesson 5: Playing Defense in the Information Wars Makes You a Primary Target of Lies, Ironically With No Ability to Counter Them
Unsurprisingly, when you fight back against liars, you become the target of lies yourself. Here are a few false claims and narratives that those playing defense, because of the rules of engagement, were unable to respond to in real-time:
Claim: Those playing defense are censoring Americans.
The accusation that those playing defense censor content from social media platforms was, and remains, false. Meetings we facilitated between the private and public sector consisted of highlighting instances where the internal rules of platforms (called “TOS” or Terms of Service) were broken. Importantly, it was not the government’s rules being broken, it was the platform’s rules. When informed of content that broke its rules, platforms sometimes took action, sometimes not.
The fact that we could be, and often were, ignored, makes the claims of Orwellian government censorship even more absurd.
There were strict rules against tracking influence operations originating in the United States or with Americans. Although important and necessary, these rules represented a massive design flaw and obstacle to doing our jobs. Hostile foreign nations laundering their talking points through domestic actors had multiple benefits: plausible deniability, language and cultural localization built in, etc.
As long as a narrative came from an American, it was deemed protected speech, and those on defense could not touch it, even if the link to foreign influence was clear and undeniable. Only if money laundering and Foreign Agents Registration Act (FARA) violations were found could these influence networks be exposed.
Claim: Those playing defense were willing to break the law to enforce it.
I guess we can blame Hollywood for this one. There’s a common misconception that people working in government care little about following rules. We saw firsthand, there is no more risk-averse group than people working in government. Most are laser-focused on a slow, steady, and non-controversial rise through the bureaucracy into their retirement. Taking any risk, even if it’s in your job description, is explicitly avoided.
Many would rather do nothing at all than do something considered controversial by anyone, which made playing defense all the more difficult.
Claim: Those playing defense seek to control information.
This is a big one, and a common point of misunderstanding. Yes, of course, there were efforts at debunking the most obvious lies and falsehoods, and promoting objective truth. But those on defense in the Information Wars quickly realized that adjudicating political nuance would ultimately prove self-defeating. Politics often operates in a gray area where nothing is quite black or white. Getting lost in those discussions would make us an arbitrator of truth, something we had no interest in.
This is why our efforts focused largely on exposing attempts at manipulation and fake behavior. Our job was to expose anyone pretending to be someone else, using a false identity. The platforms refer to this as “coordinated inauthentic behavior”. This framework also has a fatal flaw: when interacting online, fake support feels the same as real support. Do most people really care if a viral tweet was artificially inflated by bots? Your endorphin response does not, nor does your ego.
Importantly, the U.S. government has no mechanism for removing content from social media. Even when laws like revenge porn are broken, requests are still at the mercy of whether or not the platform decides to act. Which brings us to the next problem.
Lesson 6: The Platforms Just Do Not Care
To be clear, in our work, we met dozens of individuals who were intensely devoted to the work of preventing harm online. And not all companies were the same. Some platforms actively prioritized rooting out harmful content and rule violations, until they did not. Individuals tasked with rooting out harmful activity often cited their own leadership as the biggest obstruction to doing their jobs.
Sound familiar?
Platforms would often do the bare minimum to avoid accusations of doing absolutely nothing, perhaps to avoid embarrassment and negative headlines. They often took the “COVID testing” approach to foreign influence operations on their network. If they did not know about it, it did not exist, and most importantly, they were not obligated to do anything.
Social media platforms are publicly traded companies, they lack the structural incentives to care about threats to the public good. They are set up to maximize revenue via impressions and clicks. Therefore, dedication to fighting lies and manipulation is 100% conditional. If an influence operation with enough potential reputational harm was found and shared with a platform, we could be relatively certain it would be removed. However, if an influence operation did not fit this criteria, platforms would routinely ignore dangerous and malicious networks of actors.
It gets worse. In 2025, platforms have begun counting bots, trolls, and influence operations as metrics and daily active users they can sell ads to. In retrospect, this was inevitable. Clicks are clicks. If you are not working to uncover the nature of that activity, then as far as anyone knows, these are real users who can generate real ad revenue.
Instead of studying, examining, and rooting out coordinated inauthentic behavior, platforms have started to monetize it.
It gets (even) worse. In an additional attempt to increase revenue, platforms are now offering prioritized distribution of content, purchasing a wider audience for your narratives for a small monthly fee. This pay-to-play system is a no-brainer for those on offense, spending a small amount to ensure maximum exposure. Making things worse, there are no meaningful “know your customer” checks on this process, meaning anyone with a credit card can do it.
This perfect storm represents a logical endpoint to a complete lack of standards, principles, and ultimately a lack of regulation. This is why some argue social media should be a public good, even a utility. You simply cannot expect a highly profitable company to act in the public good if it means turning off, or even slowing down its money printing machine.
Lesson 7: Algorithm Transparency Was the Way Out of This, but It May Be Too Late
A simple way to limit covert influence on social media would be to make public how predetermined logic makes decisions on what information it shares with which users, otherwise known as the “algorithm”. If the algorithm was known to the public, it would make it more difficult to exploit. But companies view the algorithm as part of their “secret sauce”, important and proprietary intellectual property. And as long as the government agrees, the public will never know the extent to which platforms prioritize one narrative over another, and for what reason.
The Cambridge Analytica scandal, which received a lot of public attention, was merely about purchasing ads. But ad impressions represent a small % of total impressions on any social network. The real control over messaging and narratives lies in the algorithm, either gaming it from the outside, or controlling the rules it follows from the inside.
Decades on, we still have no laws or regulations forcing platforms to disclose why they promote certain content over others. The old standard was a “free market” approach. Platforms, as private companies, were left to maximize the amount of ad revenue they could generate. This created a steady state of reasonable moderation as advertisers would not want their money associated with extremism or harmful content. However, when wealthy partisan actors can simply purchase a platform and disregard traditional economics, this dynamic can no longer be relied upon.
Today, owners of major platforms consist of hostile foreign nations and/or partisans who almost certainly put their thumb on the scale in favor of certain narratives over others, in ways and for reasons that will never be revealed to those affected by it.
Because social media is how most people consume news and interact with the world, even small decisions regarding algorithmic information distribution will have massive outcomes and consequences.
Without clear algorithmic regulations, free and fair elections are impossible, as external and hidden influence becomes so obfuscated that it becomes unquantifiable. Political ad spend was just a small part of efforts to manipulate information flow. The golden goose for external parties attacking social media is gaming the algorithm to overload targets with an unlimited, low-cost supply of junk.
The reality is, control over digital information flow is an inevitability. Somebody will make the decisions about what we see, and why we see it. The choice is: we either have insight and awareness regarding this control, or we do not. That was the main goal for those on defense, full awareness of what information we are shown, by whom, for what reason, and ideally some basic (and objective) context regarding its reliability.
Lesson 8: The AI Fakpocalypse™ is Overstated
Generative AI was supposed to be a bombshell when it came on the scene. The truth? Fears of a disintegration of trust and truth online were realized years ago. We saw poor-quality Photoshops ignite global narrative campaigns. Easily debunked lies accompanied by obviously fake media could spread around the world as long it meets the zeitgeist of speaking the right message to the right people at the right time.
Convincing AI-generated fake media is not necessary when too many have already demonstrated the willingness to suspend critical thinking and spread false (but agreeable) stories. More realistic AI-generated content cannot break something that was already broken: our trust in what we see online.
AI-generated content aside, fears about draconian AI controlling the content we see have also already been realized in the proprietary algorithms powering your feed.
When it comes to how honest most online activity feels, there’s not much room for more powerful AI to make things worse.
Lesson 9: Animal Farm Won, 1984 Lost
For decades, there were two schools of thought regarding what future type of censorship would prevail. 1984 argued that an all-powerful government would censor important information. Animal Farm argued that when consuming information, there would be so much garbage and noise that it would not be necessary to censor anything, because no one could tell right from wrong, true from false.
The true threat to online speech in the free world is not draconian state censorship, as very little exists on Western platforms relative to the greater conversation. The far more realized threat is the flooding of spam and noise by bots, trolls, and increasingly fake and automated sources. This renders platforms we’ve built to facilitate all of our vital communication nearly unusable.
The marginal cost of creating noise online is approaching zero. Flooding the zone with noise is a far easier way of burying the truth.
Influence operations a long time ago largely stopped trying to convince anyone of anything, but rather aimed to make everyone suspicious and distrustful of everything. An electorate filled with cynical, nihilistic citizens is a compromised one, precisely the goal of those on offense.
Lesson 10: What Is at Stake in the Information Wars
Everything.
If those on offense succeed in overwhelming the digital networks we rely on for basic communication, our society becomes paralyzed.
On a macro level, if a population cannot maintain a shared objective reality, any collective action becomes impossible. How can people work together to fix something if they cannot even agree it is broken?
This is particularly bad timing as most of the world’s existential threats require collective action, such as pandemic response or climate change. Even a democracy’s ability to accurately reflect the will of the people via elections is threatened when people cannot accurately know the state of their own country and vote accordingly.
What is the point of holding an election if no one knows what they are voting on?
Losing the Information Wars means the end of peaceful self-governance and coexistence, and all that comes with it.
This is a war the free world is losing, largely because short-sighted politicians and private companies are destroying our ability as citizens to be equipped to fight it. An effective and resilient response does not align with their narrow and selfish interests.
We must fight back. Demand algorithmic transparency from social media platforms, and stop using them if they refuse.
Do not vote for politicians who disregard truth, or worse, use it as a weapon.
Above all else, we must look in the mirror. If we too often prefer a comforting lie to an uncomfortable truth, we will get the future we deserve.
The internet has become a hostile information landscape, filled with fakery, deception, and Rodents of Unusual Size (trolls). An information fire swamp we all have to navigate, increasingly without help, by ourselves.
If we tolerate the information swamp and continue to reward those who created and benefit from it, we will be stuck here forever.