This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
TikTok videos have been circulating that use manipulated audio to falsely depict anti-extremism protests as large-scale rallies in support of the AfD party. Fact-checkers confirmed that the footage was repurposed from 2024 protests, with AI-generated audio tracks added to create the illusion of pro-AfD chants. The accounts responsible for the content deleted the videos following inquiries, claiming no intent to spread misinformation.
TikTok videos purportedly depicted large crowds as chanting pro-AfD slogans by reportedly replacing the original audio with manipulated sound. The footage, originally from January 2024 anti-extremism protests, was repurposed in January 2025 to reportedly mislead viewers. Purported AI-generated or altered audio was added to create the illusion of far-right support. Fact-checkers confirmed the deception, and the accounts responsible later deleted the content, claiming no intent to spread misinformation.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to conduct large-scale disinformation campaigns, malicious surveillance, or targeted and sophisticated automated censorship and propaganda, with the aim of manipulating political processes, public opinion, and behavior.
Also classified as
The AI-manipulated audio tracks fabricated false chants to mislead viewers about the nature of the protests.
The incident involves the coordinated use of AI-manipulated audio to create political propaganda and influence public opinion ahead of Germany's federal elections.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The incident was caused by human actors intentionally using deployed AI audio generation tools to manipulate video soundtracks and spread political disinformation.
Added categories peak at 4: Severe (Epistemic Harm, Direct)
Ahead of Germany's federal elections, viral TikTok videos utilized AI-generated audio to falsely depict anti-extremism protests as massive rallies for the far-right AfD party. This incident represents a substantial information warfare and disinformation threat, leveraging synthetic media to manipulate public perception and influence democratic processes, though it lacked physical or economic impacts.
The AI system was used as a tool to perform a specific, well-defined function: generating or manipulating audio tracks to overlay onto existing video footage. It operated on direct human input without exhibiting autonomous planning or decision-making.
Risk Level 3: Limited Risk. The incident involves AI-generated content (manipulated audio/deepfakes) used to mislead, which falls under transparency obligations for deepfakes and synthetic media.
The AI was used to generate synthetic crowd chants ('east, east, east') to replace original audio in videos posted on TikTok.
The reports from Euronews and Correctiv provide clear, verified details about the nature of the manipulation, the platforms used, the scale of views, and the specific political context in Germany. The role of AI in generating the audio tracks was confirmed by audio detectors, and the creators themselves admitted to the uploads before deleting them.
None. The evidence clearly points to a coordinated political disinformation effort using manipulated media.
Technology
AI Audio Manipulation Tools
AI tools allowed easy creation of realistic crowd chanting soundtracks.
Algorithmic Amplification
TikTok algorithms promoted engagement, making fake news go viral quickly.
Inadequate Automated Detection
Platform systems failed to flag AI-altered audio before videos went viral.
Data Inputs
Manipulated Audio Overlays
Adding AI-generated chants to unrelated crowd footage misled viewers.
Lack of Contextual Metadata
Missing metadata let old footage from other events be easily repurposed.
Human Factors
Intentional Deception by Users
Accounts uploaded altered media to falsely suggest massive political support.
User Susceptibility to Fakes
Viewers accepted and shared the videos without verifying their authenticity.
Process and Methods
Weak Content Verification
TikTok lacked robust proactive verification processes for viral political media.
Delayed Fact-Checking Response
External fact-checkers like Correctiv only identified fakes after viral reach.
Management
Inadequate Risk Mitigation
Platform management failed to secure elections against AI disinformation.
Regulatory Environment
Lack of Platform Accountability
Insufficient regulatory pressure on platforms to police AI-generated politics.
AI system involved
An AI system was involved, but the reports do not identify it.
Identified by automated extraction; can differ from the alleged developers above.
