
NASK's analysis shows how artificial intelligence tools fake photos and generate false information after the shooting in Krakow.
AI-generated summary
NASK has published the analysis "Sharpened does not mean true" regarding photo manipulation using artificial intelligence in the context of the tragic events in Krakow.
Various prints on the perpetrator's T-shirt. Artificial intelligence added the details
One of the examples described by NASK are photos of the T-shirt of the perpetrator of the shooting in Krakow. After processing using artificial intelligence, various prints appeared on the clothes. One version featured a large white and green geometric pattern, another had an extensive yellow and green motif. There were also differences in the elements of the environment.
As NASK emphasizes in the analysis "Sharpened does not mean true", it is impossible to recover information from low-resolution, blurred or highly compressed material that was not recorded in the original image.
Generative AI systems can only estimate what details could be in a given place, supplementing the image with credible-looking elements, such as letters, contours, textures or background fragments, experts explain.
In the case of the shooting in Krakow, it is not only about the aesthetics of the photos. A sign on clothing generated by artificial intelligence or an emblem that does not exist in reality may become a starting point for further speculation and disinformation.
Artificial intelligence can show what the user is looking for
The appearance of the sharpened image is also influenced by the suggestions of the person using the tool. If the user is convinced that there is a specific symbol or inscription in the photo, the model can generate details that match his expectations - even if the original photo does not support such a conclusion.
The data on which the model was trained is also important. NASK points to an example of a wide Velcro on the sleeve, which appeared in many photoshopped photos of the perpetrator. This element was not visible in the original material. However, the artificial intelligence could have added it because during training it analyzed, among other things, photos of military jackets, which often feature similar Velcro.
A similar problem concerned patches and symbols on the attacker's clothing. There were blurry emblems visible in the original photos, but it was impossible to clearly determine what they represented. Meanwhile, on the modified graphics, AI created specific signs and patches that did not correspond to the actual items of clothing, differing, among other things, in shape.
One such graphic received over a million views on social media. Artificial intelligence was also used to alter photos of the attacker's weapon.
Fake graphic with the Ukrainian flag. Disinformation exploited the tragedy
NASK points out that not only altered photographs appeared in circulation, but also graphics generated entirely by artificial intelligence. One of them showed a masked man with a Ukrainian flag on his uniform and the Polish emblem in the background. The false graphic was quickly linked to the perpetrator of the shooting, falsely attributing him to Ukrainian citizenship.
Even though the original image was fictitious, subsequent versions and alterations continued to be published on the Internet.
Experts also looked at chatbot responses to questions about footage from the shooting. In one case, the AI model incorrectly recognized a modified frame from an event, considering it to be a scene from an American film.
How to recognize a photo processed by AI? NASK indicates four steps
The Institute emphasizes that not every interference with a photo using artificial intelligence is intended to mislead recipients. Some users simply want to see details better or are curious about the capabilities of new tools. The problem arises when generated items start to be presented as facts.
NASK indicates four basic principles that help verify the authenticity of materials:
Check the source of the photo or video. Look for the earliest available publication and determine whether the material comes from a reliable source.
Compare different versions of the material. Check whether new inscriptions, symbols, faces, background elements or lighting changes appear in subsequent publications. Details that could not be seen in the original painting should be particularly suspicious.
Separate what you see from what you interpret. A blurred shape is not evidence of the presence of a specific symbol, and the illegible inscription makes it impossible to determine its content.
Do not re-share forged materials. Reproducing graphics without checking their origin may contribute to the spread of false information.
“In crisis situations, caution and reliable verification of information are often more important than the speed of its publication,” NASK concluded.
Shooting in Krakow. Tragic result of the attack
The attack took place on Saturday on the Main Market Square in Krakow. The 24-year-old drove his car onto the market square and then opened fire with an automatic weapon on random passers-by.
As a result of the incident, six people died and 15 were injured, including two policemen. The seventh fatality was an attacker who was shot during an exchange of fire with officers. In the apartment of the perpetrator's family in Ostrowiec Świętokrzyski, the police found the bodies of three people - the parents and sister of the 24-year-old.

Donald Trump talked to Jensen Huang about Alice Huang, suspected of spying for China. Nvidia launched an investigation, and the stock fell more than 4%. The class action lawsuit alleges misleading shareholders.

Nvidia CEO Jensen Huang supports easing restrictions on the sale of advanced integrated circuits to China and maintains close relations with Donald Trump. Chinese manager Alice Huang is said to be a target of Chinese intelligence activities. Huang admitted to adultery in a letter to employees. Nvidia's board launched an independent investigation, denying any breach of national security.

The police recorded at least 10 cyberattacks on store refrigerators and cold stores in Poland. Criminals remotely raise the temperature, destroying food. Expert Krzysztof Smaga warns against attacks on blood storage devices, logistics centers and data centers. Similar incidents occurred in France, Ireland, Sweden, Belgium, Portugal and Luxembourg.

The Norwegian intelligence agency NSM confirmed the leak of data of employees of over 190 state institutions and companies from the Amadeus reservation system. The hackers gained access to names, travel routes and passport numbers. The following were affected, among others: Norwegian, Singapore Airlines and Japan Airlines.

The Ministry of Digitization renews the certificate of the mObywatel application. To use all features, you must update to version 4.93.1 by November 16, unless automatic updates are enabled.

The article explains the differences between disinformation, misinformation and malinformation. Describes how emotions, algorithms, and AI facilitate the spread of falsehoods, impacting public health, financial security, and democratic processes.