The far-right Alternative for Germany (AfD) has developed a new software suite, driven by Google Gemini, Open AI and Anthropic’s Claude, that helps party members generate so-called “rage bait” social media postings.
Central to the new campaign is the Alternita platform, registered to the AfD’s general secretary Hans-Holger Malcomeß, which promises users social media posts “in five minutes, with your positions, in your style, your branding”.
In the case of the AfD, an undercover investigation by Germany’s Correctiv outlet revealed how, in a few clicks, a populist article can be turned into provocative posts to flood social media feeds.
Posing as an AfD regional party member, the Correctiv journalist was given an online tour of the new suite by Mario Hau, social media chief of the AfD Bundestag parliamentary party.
RM Block
The version of the software Hau demonstrated in a Zoom call automatically pulled in news feeds from AfD-friendly far-right news outlets. When given access to a test account, the Correctiv journalist uploaded an article from reactionary news site Nius about rescinding citizenship awarded to immigrants living in Germany.
The journalist asked the AI software suite to generate posts based on the story, demanding “consequential and involuntary remigration”, or forced deportations, to prevent “ethnic elections”.
The software spat out content, ready for all large social media platforms, demanding “an immediate halt to naturalisations and the consistent revocation of unlawful naturalisations to prevent ethnic voting blocs and protect our homeland. #Homeland #Identity #That’sWhyAfD.”
Left alone with a test account, the Correctiv reporters had no problems uploading an article from a neo-Nazi magazine to generate a post warning of an LGBT “rainbow dictatorship”, accompanied by an AI-generated image of a burning rainbow flag.
According to Correctiv, the software is already in use on the social media channels of AfD co-leader Alice Weidel, where it notes that not all AI-generated content posted through Alternita is flagged as such.
[ German police brace for unrest as far-right AfD edges closer to powerOpens in new window ]
Another AfD figure who appears to be testing the software is a satirical, fictional politician account of “Karl Ranseier”, self-described “federal minister for remigration”. The AI politician’s Twitter/X feed, apparently operated by AfD social media chief Hau and followed by Weidel, is filled with AI-generated images showing a “Ministry for Truth”, dark-skinned men boarding a train to a “deportation airport”.
The party says the software, launched in pilot phase this week, will be ready for full implementation next year.
The AfD is Germany’s most popular political party and is polling 40 per cent in one eastern federal state that chooses a new state parliament in September.
The apparent aim of Alternita is to allow AfD headquarters to control messaging in a subtle way without hindering its decentralised army of followers and activists generating posts that attract attention and help the AfD maintain its considerable online lead on rivals.
Hau told the undercover reporter that the software’s content foundation was the current party programme.
“The whole thing is tailor-made for the AfD,” he said, according to Correctiv. The investigative platform dug into Alternita’s source code and found that, at its heart, the AfD software suite draws on Google Gemini, OpenAI’s ChatGPT and Anthropic Claude – three leading AI engines.
The Correctiv investigative team approached the tech companies to ask if the AfD’s use of their engines to generate political propaganda and fake news was in line with their terms of use.
Google and OpenAI said they are each examining Alternita’s use of their engines while Anthropic did not reply to the Correctiv query.
