AI Deepfakes Mix Genuine Footage With Fabricated Clips

Synthesized videos blur lines between archival performance and digital manipulation to deceive online audiences.

Updated on Oct. 15, 2026 in Artificial Intelligence

Bold flat-color editorial illustration showing a singular translucent geometric mask over layered, jagged volumes, evoking the digital manipulation of deepfake media.
Meta's detection systems flagged manipulated videos misusing President Volodymyr Zelenskyy’s likeness, highlighting the growing challenge of verifying synthetic digital media. AI Illustration. Upload story photo >

Live Poll

Do you generally trust the authenticity of viral videos featuring politicians on social media?

A viral compilation video misrepresents President Volodymyr Zelenskyy by superimposing his face onto footage of dancers and comedy performances. Meta's AI detection tools flagged two of the three clips as digitally manipulated.

Why it matters

The synthesis of benign public footage with manipulated AI-generated deepfakes complicates the ability of automated systems and audiences to verify digital media authenticity. These composite videos leverage public recognition of figures to mask the origins of stolen personal content.

Meta's AI detection tool identified manipulations in two clips, where facial mapping was applied to existing video footage. Creators utilized blurring techniques across the content to conceal visual artifacts—irregularities or glitches caused by rendering errors—common in automated synthesis.

The players

Volodymyr Zelenskyy

The President of Ukraine who, prior to his 2019 election, served as a prominent television and film comedian.

Meta

The technology company managing the detection tools that flagged these clips, which operates an extensive suite of social media and advertising platforms.

The details

The compilation relies on superimposition, a technique where one image or video layer is placed over another to replace the original subject's facial features. The manipulation involved aligning Volodymyr Zelenskyy's face with the bodies of an Argentinian dancer, Pablo Acosta, and an unidentified performer from an 18th birthday celebration. To mask the digital seams of these overlays, editors applied blurring filters, which reduce high-frequency visual detail to hide stitching inaccuracies.

Timeline

  1. 2014: Original comedy footage of Zelenskyy is recorded.

  2. February 2022: Source footage of an unidentified dancer is uploaded to Instagram.

  3. June 2023: Original footage of dancer Pablo Acosta is posted.

The Tech Race

This development follows the precedent set by widespread deepfake incidents where high-profile figures are synthesized into viral media. It highlights the ongoing struggle between generative AI creation tools and the automated detection models designed to identify synthetic tampering.

Users encountering viral clips of public figures should look for blurring effects around the jawline or eyes, which are often used to conceal alignment errors. If a video appears highly performative or out of character, checking source metadata or verifying the footage against public archives can confirm authenticity.

The takeaway

The rise of composite deepfakes means that even genuine archival footage can be weaponized within a larger, fabricated narrative. Viewers should prioritize identifying the source and context of the underlying clip rather than focusing solely on the identity of the person featured.

Further reading

For broader context on synthetic media detection, visit Artificial Intelligence.

Source note: This article includes information reported by Full Fact.

Live Poll

Do you generally trust the authenticity of viral videos featuring politicians on social media?