News
AI-Generated Video Falsely Presented as Kraków Attacker Spread Online After Shooting
A 10-second clip of a man supposedly announcing the attack drew millions of views on X and Facebook. Fact-checkers say it was made with an AI model, and the Polish government is monitoring disinformation after the attack.
After the 7 November attack on the Main Square (Rynek Główny) in Kraków, a short video spread online showing a young man in a car laughing and saying he would "do something bad." According to fact-checking services cited by Rzeczpospolita, the footage was generated by artificial intelligence and does not show the attacker.
What the video showed
The clip shows a young man in a car. Looking into the camera, he says: "Today I'm going to smoke a pipe," then bursts out laughing and adds: "And I'll do something bad." Social media posts presented it as footage of the attacker from before the attack.
The material circulated on X and Facebook. One copy had 2 million views and another half a million. According to Rzeczpospolita, the video was probably created from a photo taken from a Facebook profile attributed to the attacker, which is no longer available.
How the fake was detected
Demagog reported that an analysis with the SynthID tool showed the video was created with a Google AI model. An analysis in the Hive tool also confirmed its origin. Kontakt24, as relayed by rp.pl, reported a 99.9% probability of AI use from Hive Detect, while Sightengine gave 99% for selected frames.
The analysis carried out with the SynthID tool showed that the material was created with an AI model from Google - Demagog, fact-checking service
These are assessments by detection tools and fact-checkers, not a ruling by prosecutors. The percentages are classifier estimates.
The government's response
According to cyberdefence24.pl, the Disinformation Countering Center is working in crisis mode, and identified false messages are being reported to platforms. Deputy Prime Minister Krzysztof Gawkowski warned that AI-sharpened images very often mislead. Government spokesman Adam Szłapka cautioned against very convincing-looking AI material.
The false reports included claims about a second attacker, the death of a police officer, the absence of police, and that the attacker was a soldier or a Ukrainian. Minister Siemoniak said military agencies are examining disinformation originating outside Poland. Cyberdefence24.pl notes that there is no evidence it was a single coordinated operation.
What it means
Investigators from the CBZC (Poland's Central Bureau for Combating Cybercrime) are checking whether the attacker left warning signs online before the attack. According to cyberdefence24.pl, it has not been confirmed that the services received any warnings before the attack.
The case shows how quickly after a tragedy synthetic material fitting an expected narrative can appear. The video, which according to Rzeczpospolita was probably made from a photo on a social media profile, had 2 million views in one of its copies. Checking where a recording comes from helps with verification, as do analysis tools such as SynthID and Hive, which the fact-checkers used.
