Five AI Detectors Flag Viral 'Saluting Veteran' Parade Video as Fake

A Facebook reel of a scarred veteran saluting at a Veterans Day parade fooled thousands; five AI detectors and manual review say the clip is synthetic, with drumsticks vanishing mid-frame.
By Marcus Bennett
2 min readAt a glance
- Hive Moderation rated the viral veteran video 99.9% likely AI-generated.
- The clip was posted on Facebook by the 'Danny Wilson' account on Nov. 6, 2026, five days before Veterans Day.
- Detection scores ranged from 80% (Sightengine) to 99.9% (Hive) across five tools.
- Manual review found vanishing drumsticks, awards merging into uniform fabric, and a distorting American flag.
A viral Facebook video showing a scarred veteran saluting at a Veterans Day parade is almost certainly AI-generated — five independent detection tools put the probability between 80% and 99.9%. Hive Moderation scored the clip at 99.9% likely AI-produced, while Sightengine gave it 80%.
The video was posted on Nov. 6, 2026 — five days before Veterans Day — by a Facebook account called "Danny Wilson," with the caption: "Proud to Be an American Veteran ❤️." Lead Stories, which ran the fact check, archived the post before it could be deleted.
What did the detection tools actually find?
Lead Stories tested both a thumbnail image and the downloaded video file. The thumbnail results were consistent across three image-focused tools:
- Pangram estimated a high probability the image was AI-made or AI-edited.
- ZeroGPT scored it 83% likely to have been edited with AI.
- InVID (a video-verification platform from the InVID project) put the odds at 97%.
The full video didn't fare any better. Sightengine returned 80%, and Hive Moderation returned 99.9% — about as close to a verdict as these tools get.
One caveat worth noting for practitioners: AI detectors measure probability, not proof. That's why Lead Stories paired the automated scores with manual frame-by-frame analysis, which is where the case really closed.
What glitches gave it away?
The manual review found the classic fingerprints of generative video (AI models that synthesize moving footage from text prompts). Specifically:
- A button on the man's breast pocket blends into the fabric of his uniform.
- What look like military awards merge into the uniform material rather than sitting on top of it.
- An American flag in the background distorts as it waves — the white stripes enlarge and the red ones shrink, which a real flag doesn't do.
- A girl marches at the front of the line holding drumsticks, with no drum visible on her.
- When she approaches the man to salute, the drumsticks vanish from her hands into thin air.
Lead Stories zoomed in and slowed down that final moment to make the disappearing drumsticks clearly visible.
What does this change for social media managers?
Two practical takeaways. First, patriotic-emotion content timed to national holidays is now a prime vector for AI fakes, and engagement-hungry pages will keep pushing it because it works. Second, a single detection tool is not a workflow — the defensible method here was triangulation: multiple detectors plus manual inspection of small physical details (buttons, medals, flags, props).
The discrepancy in scores also matters. Sightengine said 80% while Hive said 99.9% on the same clip, a reminder that detector outputs vary enough that a lone percentage shouldn't decide whether your team amplifies or reports a video.
Expect more of these holiday-timed synthetic clips as generative video tools improve — and expect the tells to shift from vanishing drumsticks toward subtler geometry errors that only frame-by-frame review catches.
Source: leadstories.com
Marcus Bennett
Show full bio
Market editor covering money, funding and market shifts around Social platform news and policy at Reach Brief.