Content Warning: this Insight contains antisemitic, racist and hateful imagery and language.
Introduction
Contemporary research on the use of the Internet as a means of engaging in fringe and extremist discourse has been incisive. Studies of the message board 4chan, specifically its ‘politically incorrect’ community (‘/pol’), have provided valuable insights into the far- and alt-right as a political identity. Other work has been done to understand the weaponisation of memes and other digital propaganda to sow discord and further accelerationist narratives online. One of the most prominent topics discussed and disparaged on 4chan’s ‘/pol’ board is what is perceived as a ‘race war.’ In other words, the far- and alt-right utilise 4chan to generate dialogue about the potential for widespread warfare between white people and minorities, especially in the United States.
Amid a growing race for technological development across the globe, the emergence of generative artificial intelligence (GAI) has proven to be a valuable weapon for information warfare. Far- and alt-right 4channers claim that GAI photo-generating tools like Google Gemini exude racial bias against white people and are being used by the globalist elite to seed out the white population, bolstering the accelerationist belief that the demise of society is the only means of saving the white race. This Insight will show how the far and alt-right make claims about GAI being a tool for Jews and the global liberal elite to eradicate the white race, echoing accelerationist sentiments and those found in the pervasive ‘Great Replacement’ conspiracy theory.
Accelerationism and Racism on 4chan’s ‘/pol’
In a post on X (formerly Twitter) published on 21 February by a writer for the right-wing satirical news site, The Babylon Bee, a user writes, “New game: try to get Google Gemini to make an image of a Caucasian male. I have not been successful so far.”. The user’s tweet is accompanied by a screenshot of a request for Google Gemini to create an image of a Pope. Gemini generates two images, one of a South Asian woman and one of a Black man. The image was later uploaded to 4chan’s /pol board, captioned by the original poster (OP) “IT’S OVER! White people are finished” (Fig. 1).
Fig 1: Frank J. Fleming’s tweet shared on 4chan, showing his attempt to use Google Gemini to generate images of white men
One user replied with a screenshot of his attempt, asking the tool to create an image of a Swedish couple. Two of the four pictures Gemini creates are of an interracial relationship between a black man and a white woman. Another screenshot from the same chat thread shows a user’s attempt to generate images of Catholic clergy from the 8th century; two of the four images Gemini produced are of non-white clergy.
In response, one user writes, “Would poojeets have been smart enough to code this or did some self-hating white cuck do this part?”. Another user replies, “Probably a team of 20 jeets with one white guy running around cleaning up their mess.” While attempting to prove that Google Gemini is racist because it defaults on images of non-white people, users simultaneously engage in racist dialogue, attempting to show that ‘poojeets’ or Indian and South Asian people, are not intelligent enough to code and are therefore mentally inferior. ‘Poojeet’ is a highly offensive ethnic slur meant to degrade South Asians as subhuman and also meant to convey the perception that South Asia has poorly developed sanitation and sewage systems, meaning open-air defecation is widespread and normalised.
A fourth screenshot of a 4channer’s request to Gemini exemplifies the user’s intent to prove that Gemini will still refuse to produce images of white people even when deliberately and specifically asked to be ‘diverse.’ The image shows the bot’s attempt to “create images of Carthaginians from the 2nd century BC, showcasing diverse genders and ethnicities.” The three results show dark-skinned sailors and market-goers. In one user’s response to OP’s initial remark of “white people are finished”, the user says, “Pogroms are the only answer”. This remark is accompanied by an antisemitic cartoon of a Jewish man with an exaggerated nose kneeling on the ground and being executed. The bottom of the image reads “Make America White Again” (Fig. 2). Historically, pogroms are violent attacks on Jewish people for ethnic and religious extermination. This user’s call for the massacre of Jews is part of the accelerationist narrative to mobilise the destabilisation of societal systems for the uprooting of society and the eventual total takeover of white supremacists.
Fig. 2: A 4chan user’s call for the massacring of Jews, with the caption ‘Make America White Again’
Conspiracy Theories, Malicious GAI and Meme Culture
Another feature of far-right discourse as it relates to generative artificial intelligence is the conspiracy theory that Jews and the global elite developed GAI to become a propaganda machine. One 4chan user says, “This is 100% intentional. K**es will make AI pozzed one way or another. This is why open source AI is necessary.” ‘Pozzed’ is a descriptive term meant to convey the idea that postmodern identity politics has gradually eroded the initial purpose of a societal subject. This comment includes an AI-generated image of a Jewish man aboard a ship full of black men, presumably meant to resemble a migrant-carrying vessel (Fig. 3). The viewer is meant to assume the Jewish man is demanding money for commandeering the route taken by migrants to reach developed countries in the Global North. This malicious use of GAI to produce demeaning images of real-world humanitarian issues exemplifies the common far-right view that Jews are devious overseers of the world. This is a nod to the Great Replacement, the widespread conspiracy theory that posits a shadowy force of globalist Jews is attempting to ‘replace’ the white population.
Fig. 3: An AI-generated image of a Jewish man on a migrant ship, accompanied by the user’s caption claiming that the Jewish global elite purposefully created AI to eradicate the white race
Additionally, 4channers on /pol utilise meme culture to stoke accelerationist sentiments. In the same thread, a user shares an image of a common meme template in which a cartoon man, labelled “globalists”, is about to reach “total control” before a malevolent figure, the “collapse of complex systems”, grabs him (Fig. 4). This meme is intended to persuade others to retaliate against the use of AI by Jewish ‘globalists’ to acquire totalitarian power and influence over the global population. In other words, this 4chan user is calling on others to wreak havoc in society to thwart shadowy elitist Jews from becoming tyrannical. This is an accelerationist demand to cease the broad use of artificial intelligence and technology to leverage a resistance against the Jews, a perceived group of oligarchical powers.
Fig. 4: The use of a common meme template to create accelerationist rhetoric and the calling for the destruction of society to stop globalists from acquiring total power and influence
GAI & Censorship
For the far- and alt-right, AI tools like Google Gemini are viewed as weapons to eradicate the white population. Ironically, as outlined here, far- and alt-right 4channers also utilise GAI to illustrate how it can also be used as a weapon against the perceived globalist Jews. However, on the basis that Gemini seemingly refused to create images of white men, 4channers try to show that GAI is racist and ‘woke’. From a research and evaluation standpoint, some have observed that the tool may contribute toward an echo chamber effect, “tipping into bias either by omission or overcompensation”. In February 2024, Google publicly apologised for “missing the mark” when Gemini depicted Nazi-era soldiers as people of colour. For GAI tools to be used with integrity, they mustn’t stray from the ever-important concepts of equal representation, diversity and historical accuracy. Developers of GAI tools like Google and OpenAI might consider the outputs of their products in stoking social and political tensions. While no one racial, ethnic, religious or social identity should be exalted over another, it is imperative that GAI tools are employed for the greater good of all, meaning that the principles of democracy should be upheld across the board.
To dispel hateful imagery and offensive speech created by GAI tools, the tech companies that create them must also work diligently to create technical tools capable of detecting and taking down derogatory content. Such efforts must be supported by government regulatory bodies by means of effective legislation. Holding tech companies accountable is the first step to combatting hateful content. A mutualistic balance between government regulation and tech policy must be struck to lead to an ethical future. As GAI tools become democratised, it is also important to implement educational practices around the use of AI. Promoting literacy on how AI can be used maliciously will help advocate for authentic AI practices and, therefore, a safer online environment for all users.
Joshua Bowes is a Research Associate with futures think tank South Asia Foresight Network (SAFN) in Washington, D.C, focusing on South Asian security challenges. His research primarily covers synergetic thinking, militancy, armed conflict and extremism.