The Risk of State-Sponsored AI Propaganda Bots

What ordinary people have no clue about is the rise of military-grade state-sponsored “social influencing” operations on social media right now.

I advised the US Gov on this and below I’m going to reveal how it works…

That person you don’t know well but who you friended and likes your posts a lot — probably a bot. They seem so real – they have friends (other bots) but also lots of friends you know who are real people.

Bot farms using AI to imitate people on social media in order to influence your opinion is a real thing, and a major threat to US National Security. But it’s already far too late. The damage has been done and without an administration (or social media owners) who care about the problem, it will only get worse.

As I mentioned above, in a previous life, I advised the US Government on combatting the spread of violent extremism on social media. We were particularly focused, at that time, on combating the spread of harmful ideology during the Gulf War.

We studied the technologies being used to spread harmful memes, the companies providing the technology, the tactics and strategies used, and the science behind it.

One thing we found – multiple US adversaries have huge operations designed to influence US society. I mean tens of thousands of operators working day and night on this, running hundreds of thousands of bots or more to influence US opinions on a very direct and personal level.

BOT OR NOT?

It’s very hard to distinguish bots from actual people, and in the extreme case it’s nearly impossible.

Here is the simple case:

And another example….

While the examples above show easily identifiable bots, in the political sphere the bots can be far more advanced. They don’t just give the same message over and over. They don’t have shallow profiles that are easy to identify as bots. They use extremely realistic personas with real-seeming profiles and friends. They post unique AI-generated messages.



In the example above – are they bots or not? Hard to tell. They say they are bots but their posts have human input as well. They may be a combination of bots and human operators in fact.

But while the examples above show actors that disclose “bot” in their names, and may include both human-generated and bot-generated content in their feeds, the vast majority of bots are fully automated, and the best ones (those by sophisticated state-actors) are indistinguishable from humans to the untrained eye.

CO-AMPLIFICATION

The most advanced bots are able to build engagement over time with targeted users and communities. They use a variety of techniques to accomplish this.

The usual approach is at first indirect and slow. They like your content, then they try to friend you – perhaps they appear to be a pretty girl or a professional looking guy – and they like your posts, comment on your posts etc. They also befriend your community, your circles. They create links with people you know or people you respect to increase the chances you will accept them.

There is also another technique which is more of a “swarm” approach – to massively amplify messages and surround or pepper a subject with bot content. This is a technique called “co-amplification” where bots artificially amplify the sharing and agreement among each other to create the illusion of widespread social acceptance of an idea.

Source

Source

Source

Source

Source

ENGAGEMENT AND REALITY DISTORTION FIELDS

The bots begin with simple sharing of fake news, and they then use co-amplification to increase visibility and game the algorithm to grow visibility of their posts. This can be targeted towards specific communities, issues or even to swarm and surround targeted individuals with stories that are designed to shift their perception of reality.

Gradually the bots up the game by making positive comments, and sharing content they like too. Bit by bit they create familiarity. They tell you little personal details and express sympathy or try to get sympathy from you. They share their opinions, in a way that are close to your own, only a bit shifted… they try to get you to consider their view, to see it their way, to doubt your own beliefs, to doubt the sources you trust…

Then the game moves up a level. They use not just 1 bot but several to do this, around each target person or group.

The theory here is that people decide what is real and true based on the circle of people around them and their beliefs. It only takes a few people to change someone’s perception of reality.

If say 3 out of 5 people you have a positive relationship with on social media think x, then you will probably start to think x too.

So what they do is they create a bot bubble around a target person or group, where there are enough of them to change the target’s perception of reality.

These bots start posting material that is designed to influence the target’s opinions. They agree with each other, they create a community. The target sees them agree and doesn’t want to be ostracized by them. Gradually the target’s mind changes to be more like their community (of bots around them).

They also undermine the sources of truth that that target’s present belief system is based on. They spread seeds of doubt about the media sources they consume, the people they trust or respect, and the interpretation of the news. This is very key – undermining the foundation of a person’s belief system, by causing them to no longer trust their sources or friends, in order to replace it with a new one.

Here’s where it gets even worse. They also share automatically generated fake news from very legit looking sites. And they also alter real news to shift how it will be interpreted.

This was applied very effectively during several US elections by our adversaries where they literally drowned out all major media with tens of thousands of fake media sites spreading disinfo on Facebook and Twitter. Using this technique Russia completely drowned out the real sources of truth during a previous US election.

Let’s say for example that you are a moderate Christian. But then let’s say a Christian priest or maybe a Bishop makes a sermon about having mercy toward the poor and disenfranchised. These are in fact the actual teachings of Jesus and you know this to be true. But that sermon happens to be directed towards a political leader who you support. Now you have cognitive dissonance and this is uncomfortable.

All it takes is a few dozen – or hundred – Christian affiliated bots to post comments and shares denouncing the sermon for you to start to sympathize with their sentiment. Instead of agreeing with the sermon you “see it through their lens”

Instead of your own. And voila, your uncomfortable cognitive dissonance is now solved and you feel good again. Then you share the negative sentiment reposts and the bots reward you with likes and comments – now you are a hero!

Even though the original words of the sermon are literally the teachings of Jesus, your belief system reorganizes to find reasons to criticize it. Now you feel even more Christian than before! Not only are you a hero, but now you are doing an important service by “defending Jesus” AND the political leader you affiliate with. And you know this to be true – you are in the right – because dozens or thousands of “people like you” appear to agree and give you “likes” and “hearts” for sharing your agreement. The “majority vote” of your reality bubble completely confirms this.

People are extremely susceptible to this kind of manipulation because in fact very few people are truly confident in their own beliefs and principles – they look to other people – and powerful leaders – instead of into their own hearts – to calibrate their beliefs. What they want most is to be perceived positively in the eyes of their circles – what they fear is to be alone, outcast and unpopular. By simultaneously wielding the power to give huge social reward or punishment – social influencing operations exert both positive and negative conditioning on populations.

SOCIAL PHYSICS

Belief systems have centers of gravity. It’s similar to physics. People are emotionally “pulled” by “social gravity” towards the strongest centers of influence around them. This is how humans are wired. It’s deeply ingrained. So be careful which communities you connect with.

In fact there are specific algorithms for identifying people who are susceptible, or who are easy targets. In particular the easiest targets are “persuadables” and those who are “undecided” – people who are moderate and can “see it both ways.” It’s much harder to take a person who is deeply set in their ways and shift their reality – but it’s trivially easy to manipulate “persuadables” – and in fact most reasonable people are “persuadables.”

Now it gets worse. These bots are designed to gradually increase their level of engagement with targeted people. When they get a response or engagement from a person, they try to have conversations – this is a slow approach over weeks or months. When the target engages the bot immediately cuts over to a real human operator who takes over for the actual dialogue, so the target at that point is interacting with a human not a bot.

The bots scale what a human operator can do, enabling them to groom and influence hundreds of people.

Effectively, for example, 100 operators can run 10,000 bots influencing 100,000 people or even millions.

This technology is VERY advanced. Far more advanced than anyone realizes. It is being applied at a large-scale now on Twitter, Tik Tok and Facebook. It is miltary-grade technology being aimed at you and me and everyone you know 24/7.

WHO BENEFITS?

A large portion of the posts – especially comments but also lots of sharing and news and ads and boosted posts – on Twitter, Facebook and Tik Tok are in fact state-sponsored “social influencing” operations at work. We are all, to quote the famous Palmolive commercial, “soaking in it.”

And you can bet that the owners of Twitter and Facebook are 100% aware of what is going on (and profiting immensely from it). But it’s hard to detect and hard to stop, and when it supports their political and business agendas they also are not likely to intervene. I know this is true because when I was working on this I alerted the leaderships of some of these companies and their reaction was telling. They denied it.

Yet about a year later when news came out proving it, they suddenly changed their tune (temporarily). Fact checking was the proposed solution. It worked to some degree. But now Fact Checking is ended. Fact Checking is a cost center and ultimately reduces ad revenues, and increases liability. I can see why they don’t want to do it. Yet as stewards of the new global town squares, it is a moral obligation. This of course has no meaning in their board rooms.

Who is running these operations? Russia primarily. China, Iran, and other states as well. And there are also large privately-funded operations in play as well.

For a deeper analysis read this report by the US Army War College: https://publications.armywarcollege.edu/…/understandin…/

WHERE ARE WE TODAY?

In a nutshell, we are in a worse situation than before. Now it is open season on you. On all of us. We are up against well-oiled political and social influencing machines backed by billions of dollars and military entities, and private entities with military-level technology, with the tacit and even active support of social media companies who profit from this or are aligned politically.

They cannot be stopped because they are outside of US Jurisdiction, and the new administration and social media owners turn a blind eye and have zero interest in stopping them for obvious reasons as well.

This military-grade technology is largely responsible for the current political and social upheaval in American society. They literally datamine everyone’s social media, phone records, election records, and more – they profile everyone and target voters in particular districts this way, to shift their votes.

Our societal crisis – the extreme radicalization of both the left and right against each other – was and is being engineered using military-grade technology by our adversaries. You can take this fact to the bank.

And by the way – for the moment it may appear to many on one side of the political spectrum that these memes are good because they support their beliefs and politics. But you should have no illusions if you happen to be in that camp. Our adversaries don’t care about your political party or your beliefs. They have one single long-term goal – and that is to use the very strengths of Democracy (free speech and liberty) to cause the USA to self-destruct. They want the USA and Western influence to die.

Whether you are a liberal or a right winger you should not want this to happen.

THE EPIDEMIOLOGY OF MEMETICS

Another fun fact – we studied the similarity of the spread of harmful memes (violent and radical extremist ideology for example) and we applied the science of epidemiology to measure their spread.

We found that certain ideologies – for example ISIS and Al Qaeda – had a low viral co-efficient, most similar to diseases like Leprosy. Leprosy spreads only through long-term close contact. It does not spread easily in fact. It spreads through family contact, people living in the same household, and it’s slow.

But then there are memes such as the disinfo being spread by political extremist groups in the USA today. These memes are more like smallpox – even people who show no symptoms can spread the disease to others.

The spread of harmful memes – fake news and disinfo and hate content – is a true public health emergency. It’s a mass mental health crisis. It’s an epidemiological problem.

Fortunately there are strategies from epidemiology and public health for countering this problem. The basic methodology is to use inoculation and “ring immunization” to target at-risk populations with counter-memes that either immunize them in advance or cure them once they are infected. But to do this effectively you need to have the same level of massive scale and technology that the state-sponsored adversaries have. But nobody with the means to do this has any interest in it.

While there is no large-scale countermeasure in place to counter this threat – at least being aware of what is really going on – the meta-world above the “world” we see on social media – can provide some level of immunity.

WHAT IS NEEDED?

When I advised the U.S. Government on this very issue – we found that the our government DOES have the capability to combat this yet because of U.S. laws they cannot use it towards American citizens, they can only apply it towards non-citizen populations. I can see why this law is important and very risky to repeal. Yet at the same time it means that the government is powerless to defend the U.S. population from this kind of state-sponsored activity. It’s a damned if you do, damned if you don’t situation. Is there a lesser of the two evils here? I’m not sure.

What is missing from the equation is any entity that is truly concerned with public health and is not politically aligned. There is literally nobody in this game who simply wants public mental health and safety. It’s really too bad nobody wants to fund that. I have tried to find them. I don’t think they exist.

But what I do know is that the polarization of our society is no accident and it’s also not a natural phenomena. It’s an engineered problem created and operated by adversaries of the USA. And unfortunately much of our political class is all too happy to play along while it suits their goals.

Americans (and reasonable people everywhere) are being weaponized against each other. This activity is the most “Un-American” behavior and you should not participate in it. Instead try to immunize those around you by raising awareness of what is really happening, but by all means don’t polarize yourself and others against your fellow citizens in the process – because that is playing right into the hands of those who are working to destroy us.