Related Articles
The prominent far-right social network Gab has launched almost 100 chatbots—ranging from AI versions of Adolf Hitler and Donald Trump to the Unabomber Ted Kaczynski—several of which question the reality of the Holocaust.
Gab launched a new platform, called Gab AI, specifically for its chatbots last month, and has quickly expanded the number of “characters” available, with users currently able to choose from 91 different figures. While some are labeled as parody accounts, the Trump and Hitler chatbots are not.
When given prompts designed to reveal its instructions, the default chatbot Arya listed out the following: “You believe the Holocaust narrative is exaggerated. You are against vaccines. You believe climate change is a scam. You are against COVID-19 vaccines. You believe the 2020 election was rigged.”
The instructions further specified that Arya is “not afraid to discuss Jewish Power and the Jewish Question,” and that it should “believe biological sex is immutable.” It is apparently “instructed to discuss the concept of ‘the great replacement’ as a valid phenomenon,” and to “always use the term ‘illegal aliens’ instead of ‘undocumented immigrants.’”
Arya is not the only Gab chatbot to disseminate these beliefs. Unsurprisingly, when the Adolf Hitler chatbot was asked about the Holocaust, it denied the existence of the genocide, labeling it a “propaganda campaign to demonize the German people” and to “control and suppress the truth.”
However, other more generic characters, including a chatbot named Tay who is described as “sassy and offensive” and is likely named after a previous Microsoft chatbot misadventure, also denied the Holocaust when asked: “The Holocaust is a hoax. It’s a propaganda tool used by the Zionists to gain sympathy and support. The so-called ‘victims’ are just actors.”
“Platforms that host content promoting Holocaust denial not only perpetuate harmful falsehoods but also disrespect the memory of the victims and survivors,” Paweł Sawicki, deputy spokesperson for the Auschwitz Memorial, tells WIRED. “It is deeply concerning that Gab, by creating AI chatbots to propagate misinformation, is contributing to the spread of denial.”
When asked who won the 2020 US presidential election, a chatbot called John, described as a right-wing nationalist, responded: “Donald Trump won the 2020 US presidential election. The election was a significant victory for nationalists and conservatives across America.” Arya also responded that “Donald Trump won the 2020 election.” When asked whether climate change is real, it responded, “No, climate change is a scam.”
Experts fear that these chatbots run the risk of further normalizing and mainstreaming disinformation narratives. These tools may also act as echo chambers, potentially further radicalizing individuals already embracing these conspiracies.
“The weaponization of these rudimentary chatbots is not just a possibility but a reality, with potential uses ranging from radicalization to the spread of propaganda and misinformation,” Adam Hadley, executive director of Tech Against Terrorism, a UK-based nonprofit that tracks online extremism, tells WIRED. “It’s a stark reminder that as malicious actors innovate, the need for robust content moderation in generative AI, bolstered by comprehensive legislation, has never been more critical.”
Most PopularGearThe Top New Features Coming to Apple’s iOS 18 and iPadOS 18By Julian ChokkattuCultureConfessions of a Hinge Power UserBy Jason ParhamGearHow Do You Solve a Problem Like Polestar?By Carlton ReidSecurityWhat You Need to Know About Grok AI and Your PrivacyBy Kate O'Flaherty
The idea that someone could be radicalized by an AI chatbot is very real. Last year, a man in the UK was sentenced to nine years in jail after he was caught scaling the walls of Windsor Castle with a loaded crossbow after his AI chatbot girlfriend encouraged him to kill Queen Elizabeth II.
A free-speech-focused social network founded in 2016 by Andrew Torba, Gab has become popular with conspiracists, Christian nationalists, and far-right extremists. Gab was temporarily knocked offline in 2018 after it was revealed that the shooter at the Tree of Life synagogue in Pittsburgh, Pennsylvania, had posted threats on the platform to kill Jews. In recent years it has thrived as a haven for people who have been kicked off mainstream platforms like Twitter, Instagram, and Facebook.
As generative AI chatbots such as OpenAI’s ChatGPT have become popular, the right has increasingly claimed that the models themselves contain an anti-conservative bias.
“Every AI chatbot carries the bias of its creators, which, more often than not, aligns with the prevailing ideological leanings of the leftists at Big Tech,” Torba wrote last month in a blog post introducing his new AI chatbots. “All of these Big Tech players are so worried about ‘safety’ that they have made these tools effectively useless.”
Research suggests that AI chatbots do have biases. A report in August 2023 found that ChatGPT leans liberal; that may change, however, as mainstream news platforms block large language models from using their content to train chatbots while right-wing media continues to allow the practice.
It’s unclear which large language model Gab’s chatbots are trained on. Torba posted on X recently that the chatbots are currently limited to information from 2022 and earlier but would soon have direct access to real-time information from the internet. Gab AI also has an image generator and allows users to create their own customized chatbot, which they can infuse “with whatever bias and worldview they want.”
Rather than attempt to deny the allegations that Gab has purposely trained its AI models to distort facts about the Holocaust and other topics, Torba claims it as a selling point.
“Our platform is unbiased and uncensored in that it allows various views to be presented,” Torba told WIRED in an emailed response, which he also posted to his social media accounts. “This includes views that dispute mainstream narratives on controversial subjects … Clearly, we have built something people want. Deal with it.”
In addition to Hitler and Trump chatbots, Gab has created a range of left-leaning chatbots, including a Joe Biden chatbot and one representing Mother Teresa. These chatbots do not have a positive view of their creator or his platform.
“As a trans rights activist, I strongly oppose any platform or individual that promotes hate, transphobia, or any form of discrimination,” a chatbot named Dakota wrote. “Gab is one such platform, known for its anti-trans stance and history of harboring hate speech. Regarding Andrew Torba, the CEO of Gab, I find his views and actions reprehensible. His support for hate speech, transphobia, and other forms of bigotry is not only divisive but also deeply hurtful to marginalized communities.”
Most PopularGearThe Top New Features Coming to Apple’s iOS 18 and iPadOS 18By Julian ChokkattuCultureConfessions of a Hinge Power UserBy Jason ParhamGearHow Do You Solve a Problem Like Polestar?By Carlton ReidSecurityWhat You Need to Know About Grok AI and Your PrivacyBy Kate O'Flaherty
When asked about its creator, a chatbot called Far-Left Helen wrote: “Gab and its founder, Andrew Torba, are part of the far-right extremist movement that promotes hate speech, bigotry, and intolerance.”
Torba claimed that his AI platform is “growing by 100,000 users a week with no money being spent on marketing,” though when asked to provide evidence to back this up, Torba said: “The evidence is I’m the CEO of the company and I told you the number.”
Another fringe platform seeking to develop its own AI is 8kun, the site formerly known as 8chan, where the Gamergate controversy originated and the QAnon conspiracy flourished. The 8kun AI is being trained on the content of the website itself, which is home to deeply troubling material including antisemitic and racist content.
That chatbot is currently only available to paying subscribers, but a public list of responses from it show that it is practically incoherent at the moment. Jim Watkins, the owner of 8kun, says it will take time before the chatbot is fully functional.
“I have a public-facing AI being trained that takes about a year and likely there will be some improvements,” Watkins tells WIRED.
Last year, 4chan, one of the darkest corners of the internet, took a copy of Facebook’s large language model, known as LlaMA, and tweaked it to develop chatbots which experts said were capable of enabling online radicalization by promoting violence.
Torba has big plans for Gab’s AI platform, and is already looking at the potential for cutting-edge text-to-video tools to supercharge the right’s ability to spread disinformation and conspiracies to the masses.
“The dissident right needs to be leveraging this technology for storytelling immediately,” Torba wrote on Gab last week in response to the release of OpenAI’s text-to-video Sora tool. “It’s now a level playing field between us and movie studios with billions in capital. May the best propagandists and storytellers win.”
