Image generated via artificial intelligence, using the prompt “Computer screen displaying the Turkish flag upside down” (2025) | Shutterstock AI
The ongoing protests in Turkey, triggered by the detention of Istanbul’s mayor Ekrem İmamoğlu, following the annulment of his university diploma by Istanbul University, constitute the largest cycle of political mobilization Turkey has witnessed since the Gezi Park protests in 2013. Even though the mayor was in prison, his party, the Republican People’s Party (Cumhuriyet Halk Partisi), used AI to produce a speech, ostensibly read by İmamoğlu, beamed to hundreds of thousands of people at a rally on March 29, 2025.
“I am not afraid, as I have your support,” İmamoğlu’s AI-generated voice declared. Thanks to AI, Erdoğan and his Islamicist regime had failed to silence the voice of Turkey’s most significant political opponent.
The use of AI in Turkish politics—by both the government and the opposition—is not new. In 2023, before the dual elections, the documentary platform 140journos used AI tools to prepare post-election scenarios for Erdoğan. In the scenario where Erdoğan lost, he was depicted playing tennis and spending time with his grandchildren, struggling to adapt to retired life. In the second scenario where he won, Turkey turned into a “galactic empire,” with Erdoğan portrayed in uniform aboard a spacecraft.
Not to be outdone, the Islamist opposition party, the Felicity Party (Saadet Partisi), published a video composed of Erdoğan’s famous sayings visualized by AI, under the hashtag “if anything is artificial, it is Erdoğan.” The caption read, “The images might be created by AI, but the realities are by one person.” The party also prepared a campaign song and clip using AI technologies, demonstrating that AI-assisted satire was not limited to secular or leftist actors. Across the political spectrum, opposition groups were using generative tools to craft sharper messages under the competitive authoritarian regime.
While the recent Turkish protests exhibit similarities with previous cycles, what is new is an AI-assisted protest aesthetic—a dispersed, grassroots use of image generators and language models to enhance political messaging, satire, and visual storytelling. Protesters—many of them younger and digitally fluent—used AI tools to create stylized images representing dissent. These interventions helped organizers and protesters transmit voices, communicate emotion, mobilize symbols, and imagine alternative futures in a context where direct political critique is often criminalized.
One of the most unexpected icons of this cycle was Pikachu. A video of a protester dressed as the beloved Pokémon character, likely to protect their identity, sprinting away from riot police became an internet sensation, followed by a flood of memes and AI-created images. One such image showed Pikachu being handcuffed by Turkish police, highlighting the repressive environment where even a cartoon figure could be “detained.” Another showed Superman and Joker running from police alongside a Turkish flag-carrying Pikachu—turning these global pop culture figures into symbols of injustice in Turkey.
Some AI-generated images depicted a giant Pikachu fleeing police; others showed Pikachu holding a banner reading “Release the kids,” referring to students detained for participating in protests. While “creative resistance” based on national pop culture has been a part of the repertoire of dissent, as Lisel Hintz argued in 2021, the use of global icons like Batman, Joker, Spiderman, and especially Pikachu helped the protests and their messages resonate globally. Indeed, protesters wearing Pikachu masks or costumes soon appeared at solidarity protests in Georgia and at Hands-Off events in the United States.
Protesters embraced this blend of activism and pop culture, donning superhero costumes and carrying AI-generated posters as part of their demonstrations. Resorting to what Yale professor Yil-Jan Lin (2025) calls the “subversive power of cuteness,” protesters not only mobilized international sympathy through the vulnerability associated with “cute” figures but also injected positivity and energy into a moment otherwise marked by political despair.
Especially online, this protest imagery allowed activists to maintain anonymity while still expressing outrage—crucial in a country where sharing a political image can lead to arrest or job loss. The AI aesthetic—slick, stylized, and sometimes surreal—provided a new visual vocabulary for critique and hope. The result was an aesthetic of dissent that was fluid, anonymous, and difficult to censor—perfectly suited for a political environment where even seemingly innocuous content can carry risks.
Protesters also used generative tools to imagine alternative realities. One particularly striking example was an AI-generated image of Mayor İmamoğlu depicted as a bodybuilder, with the caption, “These shoulders can carry this burden.”
Recently, there has been a renewed interest in Mustafa Kemal Atatürk, the founder of the Turkish Republic, and the early republican period, as a reaction to years of stealth Islamization in Turkey. Resorting to Atatürk and nationalistic symbols like the Turkish flag brings protesters a sense of legitimacy—and a certain level of protection—against government repression, as the regime frequently frames its own legitimacy in terms of being “national and local.”
As a result, slogans like “We are the soldiers of Mustafa Kemal,” popularized during the mass Republican Rallies of 2007–2008, have resurfaced in this new protest cycle. Given this, it is not surprising that other examples of speculative visual politics included AI-generated texts and speeches “authored” by Mustafa Kemal Atatürk responding to contemporary developments. Some content creators even generated AI videos of Mustafa Kemal Atatürk addressing protesters at rallies. These imagined futures (or past) do not assert false claims; instead, they evoke desired realities. Importantly, they offer emotional release and narrative creativity in a context where direct confrontation can be dangerous.
As the sociologist Zeynep Tufekci wrote in 2017, technological developments and enduring cultural shifts sometimes create a fresh opening for non-institutional forms of extra-parliamentary politics, while the obstacles the protestors or the opposition under mainly hybrid or authoritarian regimes face also evolve. Scholars have argued for AI “enhancing regimes’ repertoires for controlling citizens in autocracies.” The Harvard political scientist Erica Chenoweth has also recently argued that “the AI for social movements space is rapidly developing in ways that could indeed be transformative—in both promising and perilous ways.” Turkey has seen both in this recent cycle of protests and the preceding period, as it was not only the protesters using new technology. During the 2023 electoral cycle, a deepfake video has used by the government to claim a connection between the Republican People’s Party and Kurdish Workers’ Party, a designated terrorist organization. During the current cycle of protest though, reports of police use of facial recognition technologies—one of the core instruments of algorithmic authoritarianism—created widespread concern. Protesters responded by avoiding facial exposure, using gas masks, pandemic-era masks, and other coverings to protect their identities.
The ongoing protests in Turkey are rooted in political frustration, democratic erosion and in economic despair. While AI tools have not replaced traditional organizing or forms of action—many of which have deep roots in Turkish political history, they have become quiet co-authors of the moment and supported earlier forms of opposition in novel ways. This ranges from helping to transmit voices that cannot otherwise be heard, visualizing alternative futures, anonymizing dissent, and sustaining hope amidst state repression.
In addition to showcasing democratic resilience under competitive authoritarian regimes, recent protests in Turkey also highlighting how citizens in hybrid regimes can potentially use generative technologies to punch above their weight in struggles over visibility, identity, and legitimacy, despite the possibility of AI to be used to their disadvantage.
The author would like to thank Belinda Davis, Lisel Hintz, James Miller, and Jillian Schwedler for their comments on the earlier versions of this article.
