A Ukrainian Influencer's Face Was Cloned and Turned Into a Pro-Russia Account on Chinese Social Media
What happened
In February 2024, Olga Loiek, a 20-year-old Ukrainian YouTube creator focused on mental health content, found herself on Bilibili and Douyin, two of China's largest social media platforms, speaking Mandarin, promoting ties between Russia and China, selling Russian goods, and saying she wants to marry a Chinese man. She had not made any of those videos. Someone else had, using her face and voice.
The cloned videos circulated across multiple Chinese platforms, and some were watermarked with "HeyGen," a US-based AI video creation company that launched in China in 2020. HeyGen's own moderation policy is explicit on this point: users may not generate avatars representing real individuals, including public figures, without their explicit consent. The videos existed anyway, reached an audience, and Loiek found them herself, not through any alert or enforcement action from HeyGen or the platforms hosting the content.
The deepfake persona was built to perform a specific political role: a voice endorsing the Russia-China relationship, pushing Russian goods to Chinese consumers, and projecting warmth toward bilateral ties between the two countries. The choice to do this using the identity of a Ukrainian woman, someone whose nationality sits in direct political tension with that message, gave the content a dimension that the actual person had no way to control or contest from the outside.
The incident highlighted something the source reporting made plain: the ease with which identities can be stolen and repurposed on platforms like HeyGen, and the absence of effective enforcement of their own rules. Loiek's situation was not a narrow technical failure but a pattern visible across multiple platforms simultaneously, suggesting the content moved freely and without friction across distribution channels that nominally prohibit exactly this kind of use.
What the case exposes is not that AI video tools can plausibly replicate a person's appearance and voice, that capability has been documented widely. The gap is that a policy against non-consensual avatar creation did not function as a mechanism at the moment of production. There was no provable record of who submitted the likeness, what consent verification was performed, or when any reviewer on any platform first encountered the content. Without that trail, a prohibition on non-consensual deepfakes remains a written rule with no enforcing structure behind it.
Reported impact
- Affected parties
- Not publicly disclosed
- Harm type
- Not publicly disclosed
- Scale
- Not publicly disclosed
- Financial impact
- Not publicly disclosed
- Regulatory action
- Not publicly disclosed
Classification
Relevant governance controls
Governance control mapping is not available for this record.
- No controls mapped
Not publicly disclosed
Control mapping is analytical. It does not state that any control would have prevented the incident.
Sources and evidence
This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.