|
Getting your Trinity Audio player ready...
|
CLAIM: A video circulating on social media, shared by a Facebook page identified as Ghana Dawuro, purports to show a teacher at Bole Senior High School speaking out about a viral sexual misconduct video and warning people against damaging his image. The post, which attracted over 300 reactions, claimed the video was the girl’s fault.
VERDICT: False! Three separate frames extracted from the video using InVID WeVerify and analysed on Hive Moderation, a leading AI content detection platform, returned deepfake probability scores of 69.3%, 93%, and 94.8%, respectively.
All three frames flagged face manipulation as the dominant signal, with no evidence of AI image generation.
The video appears to be a real person’s footage onto which another face has been digitally superimposed, a technique consistent with deepfake face-swap technology.
The claim that the person in the video is a Bole SHS teacher speaking in his own defence is not supported by the evidence.
Background
The Ghana Education Service interdicted a teacher at Bole Senior High School in the Savannah Region following the circulation of a video on social media that allegedly showed a teacher and a student in a sexual encounter in one of the school’s science laboratories.
A statement issued by the Head of Public Relations of GES, Daniel Fenyi, on June 16, 2026, confirmed that the Service had commenced investigations into the matter and interdicted the teacher pending the outcome of the inquiry.
Copy of the statement from GES
Against this backdrop, a video began circulating on social media, attributed to “the teacher,” in which the individual purportedly speaks out, warns about reputational attacks, and suggests the girl was at fault.
Ghana Dawuro, the page that shared the video, described it as the teacher “speaking out” following the scandal. The video carries visible TikTok and Snapchat watermarks, with the TikTok account identified as @maniru.salam78.
Verification
Track 1: Manual visual analysis
DUBAWA subjected the circulating video to a frame-by-frame manual visual inspection before running it through any detection tool. The inspection identified multiple indicators consistent with deepfake face-swap manipulation.
The most immediate red flag was lip-sync desynchronisation; the movement of the lips in the video did not correspond with the words being spoken, a hallmark failure of face-swap technology when the replacement face cannot perfectly mirror the original speaker’s mouth movements in real time.
Beyond the lip sync, the inspection noted irregular and unnatural blinking patterns, with the subject’s eyes appearing to blink at intervals inconsistent with normal human behaviour. The facial boundary, particularly around the hairline and jaw, showed visible blurring and flickering, with the face appearing inadequately merged with the neck and body beneath it. This was compounded by a skin tone inconsistency between the face and the visible neck and chest area, suggesting the overlaid face was rendered at a different colour temperature from the rest of the body.
Facial movement was stiff and mechanical, with the head appearing to move independently of the face in certain frames, suggesting a “pasted” face rather than organically captured. Lighting on the face did not match the directional light visible in the background of the video, a common failure point in deepfake compositing. The teeth and mouth interior appeared blurry and unnaturally rendered in frames where the mouth was open, and the eyes presented as glassy and unnaturally smooth, lacking the depth and micro-movement characteristic of authentic footage.
Taken together, these eight visual indicators of lip-sync failure, irregular blinking, facial boundary blurring, skin tone mismatch, unnatural facial movement, lighting inconsistency, blurred dentition, and glassy eyes constitute a pattern that manual fact-checking methodology identifies as strongly suggestive of face-swap deepfake manipulation, independent of any automated tool.
Track 2: Tool-assisted analysis
DUBAWA extracted 15 key frames from the circulating video using InVID WeVerify, a tool used by journalists globally for video verification.
Three frames were independently analysed using Hive Moderation, an AI-powered media authenticity detection platform used in professional fact-checking workflows.
The results across all three frames were consistent:
Frame 1 — 69.3% probability of deepfake content; face manipulation flagged as the primary signal; AI-generated image score: 0%.
Frame 2 — 93% probability of deepfake content; face manipulation score: 93%; all generative AI tools (including MidJourney, Stable Diffusion variants, GPT, Grok, Ideogram) returned 0%.
Frame 3 — 94.8% probability of deepfake content; face manipulation score: 94.8%; AI-generated image score: 0%.
The 0% score across all generative AI categories indicates the underlying footage is not a wholly synthetic video. Instead, the high face manipulation signal across multiple frames is consistent with a real video in which a face has been digitally replaced — a face-swap deepfake.
This is a more sophisticated form of manipulation than AI image generation, and it is the technique most commonly used to create non-consensual impersonation content.
Conclusion
The video shared by Ghana Dawuro, claiming to show a Bole SHS teacher speaking out, is likely a deepfake.
Manual frame-by-frame inspection identified eight visual indicators consistent with face-swap manipulation, including lip-sync desynchronisation, irregular blinking, facial boundary blurring, skin tone mismatch between the face and body, unnatural facial movement, lighting inconsistency, blurred dentition, and glassy eyes.
These observations were independently corroborated by Hive Moderation analysis of three frames extracted via InVID WeVerify, which returned face manipulation probability scores of 69.3%, 93%, and 94.8%, respectively, with zero AI image generation signal detected across all frames, a pattern consistent with face-swap technology rather than authentic footage.
While GES has confirmed it is investigating the original Bole SHS misconduct video and has interdicted the teacher involved pending the outcome of its inquiry, no verified statement from the teacher has been issued through official channels.
The claim that the person in the circulating video is the teacher speaking in his own defence, and that the girl was at fault, is not supported by the evidence and appears designed to exploit and extend the misinformation surrounding this case.
