Fact Checkers Cyprus has published initial findings from a technical examination of the leaked video, identifying elements that can be verified as well as areas that remain unclear and require further analysis.
Speaking on the television programme Protoseleido, Michalis Sirivianos, Dean of the School of Engineering and Technology at the Cyprus University of Technology, and coordinator of Fact Checkers Cyprus, presented the results of the analysis conducted to date.
No indication of AI-generated speech in key segments
According to Sirivianos, the sections of the video in which three individuals are seen speaking show no signs of being generated or altered using artificial intelligence. The assessment was carried out using specialised image and audio analysis tools, including InVID-WeVerify, which indicate a high probability that these specific segments are authentic.
Visible edits, but limits to current analysis
He noted that certain edits in the video are clearly visible, particularly where the footage cuts abruptly without smooth continuity. However, he stressed that it is not currently possible to determine whether additional, less obvious edits exist, explaining that this would require more advanced forensic examination.
Uncertainty regarding distorted audio segments
Caution was expressed regarding segments in which individuals presented as investors are heard speaking. Due to distortion, AI detection tools are considered less reliable in these sections, meaning no definitive conclusions can be drawn at this stage.
Broader disinformation context
Sirivianos also referred to the broader work of Fact Checkers Cyprus in examining disinformation campaigns, noting that such activity has been documented across the European Union and is commonly described as a form of hybrid threat. He emphasised, however, that in the case of the leaked video, there is currently no evidence allowing for conclusions about an organised foreign influence operation.
Analysis of the “Emily Thompson” account
The analysis also examined the online account operating under the name Emily Thompson. According to the findings presented, the account displays characteristics of semi-automated behaviour dating back to its early activity. Repetitive posting patterns, indications of recycled content and inconsistencies in positions on key geopolitical issues were identified – features commonly associated with fabricated or managed online personas.
Particular attention was given to the profile images used by the account. Technical assessment suggests a high probability that the images were generated using artificial intelligence and appear across unrelated online platforms.
What remains unresolved
Sirivianos stressed that further specialised investigation is required to establish the full origin of the video, the accounts involved and any potential coordination between them. He underlined that, beyond the assessment that the main speaking segments do not appear to be AI-generated or altered, no absolute conclusions can yet be drawn.
Also read: Video exposes bribery, black money and campaign finance fraud
For more videos and updates, check out our YouTube channel
