This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
An AI-generated video depicting a police officer punching a person was shared on social media as authentic footage of police violence during student protests in France. The video was used by an activist to condemn the French government and call for international intervention. The creator of the video later confirmed it was generated using AI to shock viewers and convey a message.
A purportedly AI-generated video depicting a police officer punching a person against a wall was reportedly shared on Facebook and X as footage of police violence during student protests in France. Activist Nathalie Yamb used it to condemn French authorities and call for international intervention. The clip predated the protests, and its creator acknowledged generating it with AI. One X post reportedly received more than 650,000 views.
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 incident involved the generation and dissemination of false synthetic media depicting police brutality that misled social media users.
The AI-generated video was used by an activist to spread political disinformation on social media, aiming to influence public opinion and manipulate political processes surrounding the French government.
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 a human activist intentionally sharing an AI-generated deepfake post-deployment to mislead the public and criticize the government.
Added categories peak at 3: Substantial (Epistemic Harm, Inferred)
An activist used a highly realistic AI-generated video depicting police brutality to criticize the French government during ongoing student protests. While the video was debunked, it demonstrates the evolving capability of synthetic media to be weaponized for domestic political disinformation and social manipulation.
The AI system functioned as a generative tool used by a creator to produce a synthetic video. The report indicates the video was 'entièrement générée par intelligence artificielle' to capture attention, showing the AI acted as a deterministic generator based on user inputs rather than an autonomous agent.
Limited Risk: The system is used to generate AI-generated content such as deepfakes, which under the EU AI Act poses a moderate risk requiring transparency obligations to ensure users are informed of its synthetic nature.
The AI system was used to generate a highly realistic synthetic video of a police altercation, which was subsequently shared on social media platforms.
The report clearly identifies the origin of the video, the fact that it was AI-generated, the platform where it was shared, and the political context in which it was used. The creator's own admission on LinkedIn provides definitive proof of the AI's role, leaving little ambiguity about the nature of the incident.
None. The evidence that the video is synthetic and was used for political influence is clear and uncontested.
While the specific AI generation tool is unknown, this does not prevent confident classification of the incident's dynamics and harms.
Technology
Optional Audit Settings
Flock made critical deterrent features like case numbers optional for agencies.
Lack of Real-Time Alerts
The system lacked automated proactive alerts for suspicious repetitive queries.
Broad Search Capabilities
Features like vehicle fingerprinting allowed precise tracking without plates.
Data Inputs
Vague Input Justifications
The system accepted unverified one-word inputs like 'investigation' for searches.
Massive Tracking Database
The system records 20 billion plate scans monthly, creating dense location data.
Unrestricted Data Sharing
Data sharing across state and agency lines expanded the scope of surveillance.
Human Factors
Abuse of Authority
Rogue officers leveraged their positions and system access for personal stalking.
Victim Fear of Retaliation
Victims hesitated to report because abusers possessed weapons and legal authority.
Unwarranted Operator Trust
System designers assumed all authorized users would act ethically without checks.
Process and Methods
Lack of Regular Auditing
Departments rarely reviewed search logs to detect unauthorized officer queries.
No Warrant Requirement
Officers could search any plate instantly without judicial review or a warrant.
Flawed Reporting Paths
Victims struggled to report abuse since the police were the ones investigating.
Management
Growth Prioritized Over Safety
Flock prioritized rapid market adoption over implementing strict user controls.
Poor Departmental Oversight
Police leadership failed to establish and enforce strict database access rules.
Ignoring Victim Complaints
Some departments dismissed early warnings from victims regarding officer stalking.
Regulatory Environment
Lack of Federal Laws
No federal legislation exists to govern or restrict the use of ALPR systems.
Patchwork of State Laws
Inconsistent state laws left regulatory gaps, making prosecution difficult.
Inadequate Legal Penalties
Some states do not treat license plate reader misuse as a standalone crime.
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.
