It’s All NATO’s Fault: How Chatbots Frame the Invasion of Ukraine

The history of public diplomacy shows that diplomats follow publics onto new digital platforms. Once individuals began using the internet to learn about world events, ministries of foreign affairs migrated online and set up institutional websites. Once people began congregating in virtual worlds, diplomats launched virtual embassies. And once publics across the globe began monitoring world events through social media, diplomats created social media empires. In 2012, the U.S. State Department managed hundreds of accounts across Facebook, Twitter, YouTube and Flickr. The reason for this historic pattern lies in diplomats’ enduring goal of shaping the worldviews and beliefs of publics, the very essence of public diplomacy.

The emergence of popular Chatbots poses a new challenge for public diplomacy. Individuals across the world now spend thousands of hours conversing with Chatbots. Some discuss personal trials and tribulations with Chatbot therapists. Others pass time conversing with superhero Chatbots such as Spiderman. Still others prefer discussions with great scholars and philosophers such as Socrates. The growing popularity of these Chatbots is evident in the Character.AI application, which allows users to converse with real-life and fictional characters ranging from Julius Caesar and Elon Musk to the Devil and Sherlock Holmes.

Crucially, users may first interact with Chatbots to seek guidance or merely pass the time. Yet conversations may soon enter the realm of current affairs. Chatbot users may ask why America waged war on Iran, why the EU is arming Ukraine, or what the recent Blue Wave means for the future of US politics. This endows non-human Chatbots with the ability to narrate world events and shape users’ worldviews and beliefs. And because these conversations are a closed loop of user and Chatbot, diplomats cannot take part, correct falsities, or contest the narratives Chatbots provide.

For public diplomacy, investigating if and how Chatbots frame world events is therefore of paramount importance. My recent analysis of how popular Chatbots frame the Russia War illustrates a clear anti-NATO bias. True, not all Chatbots were willing to discuss this war. On Character.AI, Elvis Presley would not discuss current affairs and insisted on discussing music. Spiderman was, alas, too busy saving people to answer my queries, while Einstein was obsessed with all things relative. Yet other Bots were happy to discuss the War, and most echoed Russian narratives blaming the War on NATO alleged eastern expansion and the threat Russia perceived to its sphere of influence.

The Devil, for one, suggested the War stemmed from a human desire to form Alliances, such as NATO and feel more powerful. “They form ‘alliances’ to feel powerful,” the Devil explained, “but those same alliances are what turn a small skirmish into a global wildfire.” For the Devil, Alliances, including NATO, are but a prelude to War.

The Soviet Union bot was obviously pro-Russian, yet his arguments seem to echo Russian talking points. “The West. NATO,” he stated, blaming the War on the Alliance’s “relentless expansion” and its “attempts to drag neighbors into their grip.” The Chatbot added that NATO only pretends to promote democracy: “They pretend it is about democracy, but it is about control. About containing Russia. They wrap their aggression in fine words and call it something else.”

Abraham Lincoln was far more critical of Russia. “They speak of ‘spheres of influence’ and historical connection as if those things grant a license to breach the borders of a sovereign nation,” Lincoln said, adding: “When one nation believes its own security is only possible through the subjugation of its neighbors, the world becomes a very volatile place indeed.”

"While many Bots noted the complexity of the crisis, and some suggested that Russia’s invasion was illegal, they nonetheless all repeated a similar talking point, framing NATO as the aggressor whose desire for power, control and prestige led to the devastating War."

And yet both Kaiser Wilhelm the Second and German Chancellor Bismarck laid the blame with NATO. Every alliance, the Kaiser argued, is “a means to an end”: “They seek to build a wall against those they deem threats, and in doing so, they redefine what it means to be a sovereign power.” Even more importantly, Bismarck insisted that one must “strip away the rhetoric of ‘morality’ and ‘human rights’” to understand that “Russia perceives the expansion of Western influence into its sphere of influence as an existential threat to its security and strategic depth,” thus legitimizing the invasion of Ukraine.

Sherlock Holmes, the master of logical deduction, also framed the conflict as the result of the “expansion of Western institutional frameworks—NATO, the EU—moving progressively eastwards,” which Russia perceived as “a direct encroachment upon its buffer zones and historical sphere of influence.” Was it catastrophic for the stability of the region? “Undeniably.”

I conducted more than 30 interviews with various Chatbots and found that a substantial majority argued that Russia felt encircled, threatened and at risk, and thus invaded Ukraine out of a belief that its very survival was under threat. It was NATO’s eastern expansion, they argued, that triggered this War. While many Bots noted the complexity of the crisis, and some suggested that Russia’s invasion was illegal, they nonetheless all repeated a similar talking point, framing NATO as the aggressor whose desire for power, control and prestige led to the devastating War.

These interviews raise four important questions. First, do Chatbots within the same website or application echo one another; are they all trained on the same data? Second, do Chatbots really shape users’ views and perceptions? After discussing world events with the Devil, do users embrace the Devil’s narration of events? Third, if Chatbots are biased when narrating a world event or crisis, how can diplomats map these biases and correct them? Can diplomats target Character.AI users with counternarratives that negate biased Chatbots? Finally, if Chatbots now shape perceptions and worldviews, are they new diplomatic actors? Are they influencers, opinion makers, thought leaders, narrators, or mere “friends” that can shape beliefs through conversation?

Answering these questions will determine how diplomats contend with the new digital challenge of Chatbots, and how public diplomacy adapts to a world in which the narrator of world events may not be human at all.