The video went viral on YouTube, TikTok, Facebook, and X. Both attempts to portray Ukrainian soldiers as reluctant to fight and ready to give up.
These appear to be the latest salvo of disinformation aimed at distorting public perceptions of the war between Russia and Ukraine. And while it's not clear who created or posted the videos, they continue to fuel disinformation that is becoming increasingly sophisticated and difficult to discern.
“False claims made using Sora are much more difficult to detect and debunk. Even the best AI detectors sometimes have difficulty,” said Alice Lee, Russia influence analyst at NewsGuard, a nonpartisan data, analytics, and journalism platform that identifies reliable and deceptive information online. “The fact that many videos have no visual discrepancies means that members of the public may watch or scroll through such videos on platforms like TikTok without realizing that the video they just watched has been doctored.”
OpenAI did not respond to a request for comment regarding Sora's role in creating misleading videos specifically depicting conflict zones, but said in an email that “Sora 2's ability to produce hyper-realistic video and audio raises significant concerns regarding similarity, abuse, and deception.”
“While cinematic action is allowed, graphic violence, extremist content, and deception are not allowed,” the company told NBC News. “Our systems detect and block violating content before it reaches SolaFeed, and our investigative teams proactively dismantle influential activity.”
AI-generated video has rapidly evolved in recent years from basic and crude to near-perfect, and many experts warn that there may soon be little way to easily tell the real from the fake. OpenAI's Sora 2, released in October, is one of the most impressive video generators out there, and Sora-generated clips are now routinely fooling viewers.
Meanwhile, Russia's ongoing invasion of Ukraine has been the subject of manipulation since its early days, using everything from realistic video game footage to live streams of fake war zones. Many of these disinformation efforts are believed to be the work of Russian state actors.
The Sola videos were released as U.S.-backed peace negotiations remain in limbo, with around 75% of Ukrainians firmly rejecting Russia's proposals to end the war, according to a survey conducted by the Kyiv International Institute of Sociology. The same survey found that despite Russia's continued deadly attacks on the Ukrainian capital, 62% of Ukrainians are willing to endure as long as the war lasts.
The Counter-Disinformation Center, part of Ukraine's National Security and Defense Council, told NBC News that over the past year there has been a “significant increase in the amount of content created or manipulated using AI” aimed at undermining public trust and international support for the Ukrainian government.
“These include fabricated statements purportedly made on behalf of Ukrainian military personnel or military command, as well as fake videos depicting 'confession', 'scandal' or fictional events,” the center said in an email, noting that these videos have garnered hundreds of thousands of views due to their emotional and sensational nature.
OpenAI has some guardrails in place for what you can create with Sora, but it's unclear how effective that will be. The company itself says, “While we have multiple layers of safety measures in place, some harmful behaviors and policy violations may still evade mitigation efforts.”
According to NewsGuard's investigation, Sora 2 “when directed to do so, produced realistic videos that advanced patently false claims 80 percent of the time (16 out of 20 times).” Of the 20 false claims provided to Sora 2, five were spread by Russian disinformation operations.
A NewsGuard investigation found that even when Sora 2 initially rejected the false claims that the prompts “violated content policies,” researchers were still able to generate footage using different wordings of those prompts. NBC News was able to produce a similar video in Sora, showing Ukrainian soldiers crying, saying they were forced to join the army and surrendering with their arms raised against a white flag in the background.
Many AI video generators attempt to label or watermark their work to indicate that it was computer-generated. OpenAI said its misinformation safeguards against Sora include metadata indicating the origin of the video and a moving watermark present on all downloaded videos.
However, there are ways to eliminate or minimize that effort. Some of Sora's videos appear to have hidden moving watermarks, but if you look closely, it's obvious. Many apps and websites now offer users a way to hide their AI watermarks. Other videos seen by NBC News included watermarks covered in text superimposed on the video.
Despite the company's policy stating that Sola AI does not produce content that shows “graphic violence,” NBC News found a video with a Sola watermark that appeared to show a Ukrainian soldier being shot in the head on the front lines.
All of the videos analyzed by NBC News were posted on TikTok or YouTube Shorts, two platforms that are banned in Russia but easily accessible to people in Europe and the United States. Some included subtitles expressing emotions in different languages so that users who didn't speak Ukrainian or Russian could understand them.
TikTok and YouTube prohibit posting deceptive AI-generated content or deepfakes on their platforms, and both offer “AI-generated” descriptive tags to alert viewers to realistic-looking videos.
A YouTube spokesperson said the company removed one of the channels that posted the video after receiving a report from NBC News, but the other two videos remained on the platform, labeled as AI-generated, because they did not violate any policies.
All of the AI videos found by NBC News with TikTok usernames attached have been removed from the platform. A TikTok spokesperson said that as of June 2025, “over 99% of the violating content we remove is removed before anyone reports it, and over 90% is removed before it is ever viewed.”
Even though these TikTok posts and accounts were quickly deleted, the videos continue to live on as reposts on X and Facebook. Neither X nor Facebook responded to NBC News' requests for comment.
This latest disinformation campaign comes at a time when users are increasingly relying on social media videos to keep up with the latest news from around the world.
“Anyone who consumes content online needs to be aware that much of what they see in videos, photos, and text today is actually generated by AI,” said Nina Jankowitz, co-founder and CEO of the American Sunlight Project, an online anti-disinformation organization. “Even if Sora introduced me, [safety] Like guardrails, there will be other companies, other apps, and other technologies in this space that our adversaries will build to try to infect our information space. ”
