A video on social media on 28/2 depicted an Iranian Shahed-136 suicide drone crashing into a satellite communication dome at the US Navy’s 5th Fleet base in Manama, Bahrain. The incident caused a large explosion that completely destroyed the structure.

Iran’s Tehran Times newspaper published satellite images that same day, purportedly showing the US base before and after the strike, indicating severe damage. "A US radar was completely destroyed in the Iranian drone attack", the Tehran Times social media account posted.

However, the BBC reported that this was an AI-generated product, created by editing commercial satellite images of the US base in Bahrain from 2/2025. News agency AFP made a similar assessment. One sign of the deception was that many cars remained parked in the same spots in both the original and edited images, despite nearly one year passing between the two photos.

Nevertheless, the image quickly spread across social media, garnering millions of views. This demonstrates the increasing difficulty users face in distinguishing real from fake on platforms saturated with AI-generated information.

Brady Africk, an open-source intelligence researcher in the US, stated that the number of fake satellite images often increases during major events, including the Middle East conflict.

AI-generated image of "destroyed US radar" posted on 28/2. Photo: X/TehranTimes79

"Many edited images bear typical hallmarks of AI-generated content, such as unusual angles, blurred details, and features that do not match reality. Some images appear to be manually edited by compositing damage onto real footage," Africk said.

Social media users on X last week also shared an AI-generated satellite image. It depicted a supposed US-Israel attack on an F-4 fighter jet drawing on the ground in Iran, while the real aircraft had been evacuated to safety.

One piece of evidence pointing to the deception was the insertion of nonsensical coordinates. AFP also detected a SynthID watermark, commonly used to identify images generated by Google AI, within the image.

These fabricated images appeared alongside numerous fake open-source intelligence (OSINT) analysis accounts. The activities of these accounts are negatively impacting genuine analysts.

"The fog of war makes it difficult to ascertain the success of attacks. Open-source intelligence has become a solution, including using commercial satellite imagery to verify information. However, this method is being exploited by various parties to spread misinformation," said Tal Hagin, a conflict information analyst. "Fog of war" is a term referring to the ambiguity or chaos that arises in conflicts, where information is often incomplete or misleading.

AI-generated or edited satellite images previously appeared in the Ukraine conflict and the India-Pakistan conflict last year.

"Like other forms of misinformation, manipulated satellite images can have real-world consequences if people act on information they read without verifying its authenticity," Africk warned.

AI-generated satellite image of "US-Israel attack on Iran aircraft drawing" posted on 6/3. Photo: Facebook/Sumon Kais

Satellite images still offer significant benefits in the current era, provided they are not fabricated products. "In the AI era, high-resolution, real-time satellite images can provide policymakers with crucial clues to assess security threats, as well as expose misinformation from unverified sources," AFP stated.

Following a rebel attack on Niamey airport in Niger last month, some images purportedly showing a fire at the airport's main civilian terminal were circulated online. Satellite images captured by US-based company Vantor helped determine that the material was fake and almost certainly AI-generated.

"When satellite images are used as visual evidence in conflicts, they can easily influence how people interpret events. As AI-generated images become increasingly convincing and harder to distinguish, people need to approach such content with caution," said Professor Bo Zhao at the University of Washington.

Pham Giang (According to AFP, CNN)