X Officially Admits Top Accounts Are Stealing Videos at Scale, Launches New Tools to Protect Original Creators

X Speaks Plainly: Top Accounts Are Riding on Stolen Videos

A recent post by Nikita Bier, X's (formerly Twitter) head of product, rarely pulled back the curtain on the platform's content ecosystem. He stated bluntly: "The videos on many top accounts are actually stolen from other users—sometimes only after the original video went viral five years earlier."

These remarks drew widespread attention not only because they came from a platform executive, but also because of the weight of the data behind them. According to Bier, video content currently accounts for nearly half of all impressions on X. In other words, video has become the backbone of X's traffic, and a substantial portion of it is built on unauthorized reposting.

This phenomenon exposes a longstanding structural contradiction on social platforms: the recommendation algorithms of mainstream social platforms all use "engagement signals" (likes, comments, shares, completion rate) as the core basis for content distribution, and engagement signals naturally favor content that has already been validated by the market—that is, videos that have already gone viral on other platforms. By reposting these "pre-validated hits," content thieves replicate the original content's viral potential at virtually zero creation cost, gaining a systematic advantage in the algorithmic game. Academics summarize this phenomenon as "the tragedy of the commons in the attention economy": each individual's reposting behavior is rational, but when everyone does it, the entire creative ecosystem is eroded. Facebook, Instagram, and TikTok have all faced similar dilemmas, and their responses have varied in emphasis—from technical interception and account bans to adjusting distribution weights—but no platform has claimed to have completely solved the problem. Algorithms only reward "content that generates engagement," without distinguishing between original and reposted work. For opportunists, reposting a proven viral video is far more cost-effective than creating one themselves. Over time, genuine creators end up at a disadvantage in traffic allocation.

Why the Platform Chose to Publicly Admit It Now

It is inherently unusual for a platform executive to proactively admit large-scale content theft within their own ecosystem, and the motivations behind it warrant closer examination.

On one hand, since Musk took over, X has launched a creator revenue-sharing program in which ad revenue sharing is directly tied to an account's impressions and engagement. The core logic of this mechanism is to return a proportion of the ad revenue generated when subscribers (X Premium) view creator content to the account holder. The original intent was to attract high-quality creators to stay on the platform, following logic similar to YouTube's Partner Program (YPP) and TikTok's Creator Fund. However, the structural flaw of "revenue sharing based on impressions" is that it does not distinguish content originality—as long as a video generates a large number of views, reposting accounts and original creators enjoy the same eligibility for revenue sharing. This means video theft is no longer just "leeching off the original creator's popularity," but literally taking away the hard cash that should belong to the original creator—the platform's incentive mechanism has inadvertently provided financial subsidies to the reposting ecosystem. When reposting accounts collect revenue shares based on stolen videos, the fairness of the entire incentive mechanism comes into fundamental question.

On the other hand, since video accounts for nearly half of all impressions, its copyright compliance directly affects the platform's advertising value and legal risk. Brands are unwilling to have their ads appear next to pirated content, and original creators' demands for their rights may escalate into legal disputes. By choosing to publicly admit the problem while launching new tools, X is to some extent adopting a "self-disclose first, then solve" PR strategy—getting the unpleasant part out of the way first, and paving the way for subsequent reforms.

What New Video Tools Will X Launch

According to Bier's statements, X is rolling out a series of new video-related features. Although the details disclosed so far are limited, judging from the way the problem is defined, the core goals of these tools should include the following directions.

Content Provenance and Original Content Identification

To solve the problem of "still being reposted five years later," the platform first needs the ability to identify a video's original source. This typically relies on Video Fingerprinting technology—generating a unique identifier from features such as the visual frames and audio track to match duplicate or highly similar versions across massive amounts of content.

The principle of video fingerprinting is similar to human fingerprints: even if a video has been compressed, cropped, watermarked, color-adjusted, or even mirror-flipped, as long as the core visual information has not been completely destroyed, the algorithm can still identify a match with the original version with extremely high accuracy. On the technical implementation side, mainstream solutions extract Perceptual Hashes or Locality-Sensitive Hashing (LSH) frame by frame, combined with audio spectrum fingerprints for dual-track comparison, ultimately completing massive database matching within milliseconds. YouTube's Content ID system, launched in 2007 and fully opened in 2012, is the most commercially successful example of this technology, currently covering over 100 million reference files and intercepting millions of duplicate uploads daily—however, building this system took several years and hundreds of millions of dollars. If X were to build similar infrastructure from scratch, it would face considerable technical barriers and financial pressure.

Distribution and Monetization Tools for Creators

Since video already contributes nearly half of all impressions, X has ample motivation to keep original creators on the platform. More comprehensive upload, editing, analytics, and monetization tools would not only lower the barrier to creation, but also help X narrow the gap in its competition with video platforms like YouTube and TikTok.

A Fairer Traffic and Revenue-Sharing Mechanism

If impressions and revenue could be directed more toward original content, it could theoretically shrink the arbitrage space for reposting accounts. However, such mechanisms are extremely difficult to implement—they must both accurately determine original ownership and account for growth opportunities for new accounts. Any misstep could inadvertently harm legitimate users.

The Challenges and Points to Watch in Governing Pirated Content

Admitting the problem is only the first step; the real challenge lies in execution. Content theft governance is a common problem for all UGC platforms. Even the well-resourced YouTube has long faced criticism over its Content ID system for false positives, abuse, and cumbersome appeals processes.

X's situation is even more complex. After Musk completed the acquisition in October 2022, the platform underwent one of the largest single corporate layoffs in the history of social media—total headcount plummeted from around 7,500 to fewer than 1,500, a reduction of over 80%. The most heavily impacted departments included the content moderation team, the Trust & Safety department, and the infrastructure engineering group. Copyright enforcement and content provenance systems require not only substantial engineering resources for development and maintenance, but also human review to handle edge cases and appeals processes—and these two capabilities happen to be X's weakest links after the layoffs. Meanwhile, X's data center infrastructure was also significantly scaled back during the layoffs, which poses an additional technical challenge for a fingerprinting system that needs to process massive video comparisons in real time. In addition, X has consistently emphasized "free speech" and a low-intervention philosophy, so how to combat piracy while avoiding accusations of over-censorship is a delicate balancing act.

A deeper issue is this: the fact that video accounts for nearly half of all impressions itself demonstrates the platform's heavy reliance on video traffic. Under such reliance, any governance action that might reduce total impressions will face real pressure from growth metrics. Whether the platform is willing to sacrifice some short-term data for the long-term health of the content ecosystem will be the core benchmark for testing the substance of this reform.

Conclusion: Can a Belated Admission Bring Real Change

X executives publicly admitting that top accounts are stealing videos, and directly linking it to the staggering figure of "nearly half of all impressions," is a rare moment of transparency. It reflects both that video has become the core content format of the X platform and exposes the severe imbalance in the original content ecosystem under a traffic-above-all logic.

Whether the newly launched video tools can truly turn things around depends on whether X is willing to get serious about technical investment, mechanism design, and trade-offs of interests. For the vast community of original creators, this may be a belated but welcome signal; and for the entire social platform industry, "ensuring original creators receive the rewards they deserve" remains a long-term proposition with no perfect answer.

Key Takeaways