This investigation was produced by DUBAWA and is co-published by DAIDAC.
A collage of AI-generated videos used in the Facebook posts. Photo Source: Facebook.
In recent months, Facebook has witnessed an increase in videos claiming to show bandit attacks, villages set ablaze, military operations and kidnappings across Nigeria.
Many of these videos appear authentic, attracting thousands of views, comments, and shares from users who believe they depict real security incidents.
DUBAWA began investigating these videos after a Facebook page, Emmanuel Okechukwu Aja, posted a video (archived here) claiming bandits had razed a community in Kaduna State. Initial verification revealed the footage was generated using Artificial Intelligence (AI) and was not an actual incident.
DUBAWA found the page among several accounts that repeatedly publish AI-generated videos depicting fictitious security incidents across Nigeria.
The findings raise concerns about how generative AI is increasingly used to exploit public anxiety amid the country’s insecurity challenge.
To understand the trend’s extent, DUBAWA examined three Facebook pages that regularly publish security-related content. These accounts stood out for their repeated engagement with security-related narratives, how often they shared these claims, and their audience engagement.
The investigation involved analysing AI-generated videos published by the pages, reviewing engagement metrics (views, comments, and shares), conducting reverse-image searches, and applying AI detection techniques to selected videos.
A Facebook Account Built a Large Following Around AI-Generated Security Videos
On July 7, 2026, a Facebook account named Emmanuel Okechukwu Aja posted a video (archived here) claiming that bandits had set a community in Kaduna State ablaze. The video, accompanied by a dramatic caption, portrayed burning buildings and fleeing residents, suggesting that the attack had recently occurred.
As of Aug. 8, 2026, the post has gained traction on the platform, attracting over 10,000 views and generating six comments, some of which appeared to accept the footage as genuine.
Evangelist Olufunke Aiyedehin commented, “God, please destroy all these destroyers in Jesus’ mighty name.”
Another user, Alapotiowo Thomas, also commented in doubt of the footage, “How can you people keep on imagining evil every day? As you thinketh in your heart, what you imagine is what you will get, alone not with your village people.”
DUBAWA analysed the video using Hive Detect, an AI content detection tool. The tool assessed the 10-second footage as 99.6% likely to be AI-generated, while assigning a 0% probability to AI-generated speech. The tool’s analysis timeline also flagged the footage as AI-generated throughout the duration examined.
Screenshot of the result obtained from HIVE.
Beyond the AI analysis, DUBAWA found no credible reports from security agencies, local authorities or reputable news organisations confirming that such an attack had occurred in Kaduna State on or around the date the video was posted.
DUBAWA also examined the account’s broader activity.
The account has more than 69,000 followers and lists Asaba, Nigeria, as its location. The account’s profile information revealed the user joined Facebook on Jan. 30, 2026.
Screenshot of the user’s profile.
As of Sept. 9, 2026, DUBAWA reviewed 23 videos published by the account and found that each displayed characteristics consistent with AI-generated content. Most of the videos portrayed alleged security incidents, including attacks on communities, villages set ablaze, and other violent events purportedly occurring in different parts of Nigeria.
Table 1: Some of the AI-generated videos identified on the Facebook account
| No. | Date | Link | Brief description of the content | Total views |
| 1. | 09/09/26 | https://web.facebook.com/share/r/1Sj8bedAaK/ | Video claiming bandits attacked a church in Lagos | 884 |
| 2. | 08/09/26 | https://web.facebook.com/share/r/1CT2VeptUE/ | Claim soldiers rescued victims from Nigeria | 680 |
| 3. | 07/09/26 | https://web.facebook.com/share/r/1EyY2NrTn5/ | American soldiers in the forest to fight off bandits in Nigeria | 3,600+ |
| 4. | 06/09/26 | https://web.facebook.com/share/r/14sz1mvHRa1/ | American soldiers landed in Nigeria to fight bandits | 6,000+ |
| 5. | 05/09/26 | https://web.facebook.com/share/r/19Mbycfgki/ | Bandits attacked a village in Lagos | 3,000+ |
| 6. | 04/09/26 | https://web.facebook.com/share/r/19UpGzm3Vd/ | Bandits attacked a village in Lagos | 2,600+ |
| 7. | 01/09/26 | https://web.facebook.com/share/r/14ofU2jPTtN/ | Bandits attacked travellers in Lagos | 2,800+ |
| 8. | 31/08/26 | https://web.facebook.com/share/r/19K8pRrYsN/ | Bandits attacked a church in Lagos | 14,000 |
| 9. | 30/08/26 | https://web.facebook.com/share/r/1QJQZHvtxE/ | Bandits attacked a village in Lagos | 2,000+ |
| 10. | 25/08/26 | https://web.facebook.com/share/r/1Db5Gcmp5m/ | Bandits attacked villagers in Nigeria | 3,200+ |
| 11. | 24/08/26 | https://web.facebook.com/share/r/1CHYZRMebg/ | Bandits attacked a bank in Nigeria | 887 |
| 12. | 22/08/26 | https://web.facebook.com/share/r/1GzuG9hUAt/ | Bandits attacked a marriage ceremony | 4,000+ |
| 13. | 21/08/26 | https://web.facebook.com/share/r/1GmuCB6bzF/ | Bandits attacked a courtroom in Nigeria | 4,400+ |
| 14. | 18/08/26 | https://web.facebook.com/share/r/14hY8huTSTd/ | Bandits attacked church in Lagos | 4,600+ |
| 15. | 17/08/26 | https://web.facebook.com/share/r/1BvUHGex2h/ | Bandits attacked travellers in Lagos | 4,900+ |
Screenshot of some of the AI-generated videos published on the account.
The investigation further found that Lagos State was the location most frequently mentioned in the account’s captions. However, no credible reports support many of the claims made in the videos.
The account has also built a sizable audience around this content. Its most-viewed video attracted more than 257,000 views, while the highest-performing post recorded 159 shares and 164 comments, indicating that the synthetic videos reached a significant number of Facebook users.
Peterking Stories uses AI to make a wanted bandit leader appear to speak
Another Facebook page, Peterking Stories, uses AI-generated content to portray alleged security incidents and make a real individual appear to deliver statements they may not have made.
On May, 26, 2026, the page published an AI-generated video purporting to show a bandit leader boasting about their achievements and challenging security forces. As of Aug. 9, 2026, the post had generated 580 likes, 45 comments, and 153 shares.
Screenshot of the page.
The account was created on Apr. 6, 2023, and lists Nkwerre, Anambra State, as its location. DUBAWA reviewed the page and found several videos depicting alleged military operations, bandit attacks, and confrontations between armed groups and security forces.
The page initially named “Peterking’s World” was changed to “Peterking Stories & fact” on Jan. 22, 2024. By May 23, 2024, it was again changed to “Peterking stories”.
Among the posts reviewed were videos published here and here featuring Isiya Kwashen Garwa, a wanted bandit leader. In the video, Garwa challenged the Nigerian Air Force.
DUBAWA traced the video to an image from a peace accord in Katsina, reported by Zagazola Makama on Sept. 14, 2025. This shows the image existed months before the page transformed it into the AI-generated videos it published.
DUBAWA analysed the 22-second video using Hive Detect. The tool rated the footage at 95.4%, meaning it is likely AI-generated, and a 0.8 probability of containing AI-generated speech.
Screenshot of the result obtained by HIVE.
The page repurposed an authentic image and used AI to create a synthetic video in which the person appears to speak.
Screenshot of the speech result obtained from an AI voice detector
The result supports DUBAWA’s finding that the footage was synthetically generated rather than an authentic recording of the alleged bandit leader speaking.
Table 2: The AI-generated videos identified on the Facebook account
| No. | Date | Link | Description of the content | Total views |
| 1. | 20/04/2026 | https://www.facebook.com/share/r/1Cc9J1fn9n/ | Bandit leader Isiya Kwashen Garwa challenges the Nigerian Air Force | 87,000+ |
| 2. | 19/04/2026 | https://www.facebook.com/share/r/1Fi1QkUBo1/ | The same bandit leader challenging the Nigerian Air Force | 165,000+ |
The table documents the 2 AI-generated videos of the bandit leader on the account.
AI-Generated Video Falsely Shows US Troops Heading to Nigeria
Another Facebook page, MilitaryK-9, with more than 94,000 followers, published a video on Nov. 18, 2025, that appeared to show American military personnel aboard an aircraft.
Screenshot of the Facebook page. Source/Facebook
The accompanying caption claimed that the troops were on their way to Nigeria to protect Christians. The post attracted more than 937,000 views, over 1,100 comments and more than 1,000 shares.
DUBAWA observed that the video had the Sora watermark, a cartoon-eyed cloud logo, suggesting it may be AI-generated. We further analysed the 10-second video using Hive Detect, and it assessed the footage as 99.7% likely to be AI-generated. At the frame level, Hive Detect assessed a selected frame as 99.9% AI-generated, identified characteristics associated with Sora 2, and assigned the model a 99.7% score.
Screenshot of the result obtained by HIVE.
The video was published amid heightened discussions about possible US military intervention in Nigeria. On Nov. 1, 2025, U.S. President Donald Trump said he had ordered preparations for possible military action in Nigeria, citing alleged persecution of Christians.
However, DUBAWA found no evidence that the footage showed American troops travelling to Nigeria as claimed. On Nov. 19, 2025, a day after the Facebook post, the Nigerian government said the country did not need foreign troops but sought targeted support from the US in areas including intelligence cooperation, technology, and military equipment.
The post’s timing coincided with a genuine debate over possible US intervention, making it more believable. The AI-generated footage, however, falsely gave visual form to a military deployment that had not yet been established.
The account was created on Feb. 5, 2024, with the initial name Belklef Zamani Comedy, and changed to MilitaryK-9 on Jan. 14, 2026. It lists Manchester, United Kingdom, as its location, and Nigeria as its primary location.
Screenshot of the page transparency. Source/Facebook
DUBAWA’s review found that much of its content consists of AI-generated videos depicting military and conflict-related events, including alleged Nigerian military operations, bandit attacks, bomb blasts, and developments involving the Iran-Israel conflict and the US. We observed that recent posts carry the ‘AI content’ tag.
Some of the videos reviewed attracted even higher engagement, with one recording more than 966,000 views.
Table 3: The AI-generated videos identified on the Facebook account
| No. | Date | Link | Description of the content | Total views |
| 1. | 18/11/25 | https://www.facebook.com/share/r/1BvVoRr6RX/ | American soldiers on their way to save Nigerian Christians | 937,000+ |
| 2. | 09/1125 | https://www.facebook.com/share/r/1BpDwomMWs/ | Nigerians welcoming American troops | 966,000+ |
The table documents 2 AI-generated videos claiming American soldiers in Nigeria on the account.
Screenshot of some AI-generated videos from the page
Pattern Across the Pages
DUBAWA found a recurring pattern across the three Facebook pages: AI-generated or manipulated videos were used to portray security and conflict-related events as real. The content ranged from alleged bandit attacks and military operations to international conflicts.
While the investigation did not establish that the pages were coordinated, their activities show how different accounts can use AI tools to produce and circulate convincing false security narratives.
Silas Jonathan, a disinformation analyst and the Digital Investigations Manager for the Digital Technology Artificial Intelligence and Information Disorder Analysis Centre (DAIDAC), said AI-generated security videos are effective because they exploit people’s tendency to trust what they see while tapping into existing fears about insecurity.
He said emotionally compelling, plausible footage can spread before users verify it, and social media amplification can increase its reach.
“Repeated exposure can distort people’s perception of the security situation, making insecurity appear more widespread or frequent than it actually is,” Silas said.
He warned that persistent exposure to fabricated content could also undermine trust in authorities, journalists and fact-checkers, while creating an “everything could be fake” problem in which people begin to doubt both false and genuine information.
AI Expert Warns Against Trusting Realistic Videos
Sharfadi Bature, an AI expert, cautioned social media users against relying on how realistic a video appears, saying AI-generated footage is becoming increasingly convincing.
He advised users to verify a video’s source, location, and timing, and to cross-check the claim with credible media organisations, security agencies, and independent sources. He also recommended looking for inconsistencies such as unnatural movements, distorted faces, changing objects, unusual shadows and incorrect military insignia.
“The major challenge is that seeing is no longer enough to establish authenticity,” Sharfadi said.
He added that fabricated videos can cause panic, inflame tensions, and damage reputations, while the growing prevalence of AI-generated content could also make people dismiss genuine footage as fake.
His statement echoed Silas’s advice to verify content before sharing. The disinformation analyst urged journalists and fact-checkers to focus on provenance and corroboration rather than relying solely on AI-detection tools.
“The key is to verify the claim and context, not just the pixels,” Silas said.
Conclusion
DUBAWA’s investigation found that multiple Facebook pages repeatedly publish AI-generated videos depicting fictitious security incidents across Nigeria.
Rather than isolated instances of misinformation, the findings point to a broader pattern in which synthetic media is used to exploit public concern over insecurity and drive engagement on social media.
This fellowship is implemented by the Centre for Journalism Innovation and Development (CJID) with support from the European Union.
