Why Can't AI Solve the Comprehensive Problem of 'Original Content Protection'?
-
Original content protection is an increasingly important issue for all professionals in the writing industry. However, due to various factors, it remains a problem with more talk than action, making it difficult to provide concrete evidence. With the advancement of AI technology, can this complex issue of original content protection be resolved?
At a time when wallets are taking a hit, most people, aside from receiving packages, are likely taking a rare break from their usual indulgences. So, let’s discuss a relatively serious topic—like the sudden death of a small business owner just before Double Eleven.
Shen Wenjiao, an outstanding Chinese designer who won the Red Dot Design Award eleven times in six years, collapsed at his desk on November 10. This news didn’t even make it to the trending list, with only a few people speaking out for him. He could have achieved fame and success with just one award-winning design—the Nude clothes hanger—but after countless instances of copying and imitation, two years of fruitless legal battles left his store deserted while knockoffs flourished online, ultimately leading to his factory’s bankruptcy and unpaid wages.
Shen Wenjiao’s article, Originality Is Dead, exposed the plight of original design in China. But two years later, the voices calling for change were fleeting, while infringements remained rampant. No wonder some influencers bluntly stated on social media—China doesn’t deserve originality.
Of course, such sweeping statements are frustrating for the general public. Both in theory and morality, the public and various platforms support and encourage originality. But the obstacles to social progress continue to trample over the efforts of creators.
Is there a way to armor them? Perhaps a single article can’t provide the ultimate answer, but we believe it’s time to raise this question and strive toward solving it.
Those familiar with us might wonder if we’re about to wield the mighty tool of AI again. Indeed, over the past year, we’ve heard about AI’s explosive applications in face-swapping, voice-changing, and even writing short essays. Can’t it help with originality?
Today, we’ll take a contrarian view and discuss the limitations of AI in protecting originality.
Currently, whether through government-managed copyright centers or third-party copyright websites, creators can obtain a Copyright Certificate for a fee. However, even with solid evidence, gathering proof of infringement remains a daunting task.
Works are photographed at exhibitions, images are stolen online, and some even plagiarize original content after watching creators’ live streams. More commonly, many designers aren’t even aware their work has been copied. For example, the Forbidden City’s cultural products were imitated during their crowdfunding phase and sold online before the originals.
Can machine learning solve this problem? The answer is: Yes, but it’s not worth it.
While image recognition algorithms can quickly find highly similar images in the vast ocean of online data, determining whether their use is legal or constitutes plagiarism involves complex judgments. Conducting full-scale comparisons for every work is prohibitively expensive, requiring massive server resources and impractical efficiency.
So, expecting AI systems to provide real-time alerts is unrealistic. The current state of “no report, no official action” will likely persist for a while.
Copyright certificates can curb direct copying, but most creators face something even worse—“derivative plagiarism.”
Under existing intellectual property laws, protection often extends only to “expression, not ideas.”
For example, Chanel pioneered the little black dress, but it can’t stop other brands from making their own versions—it can only prevent exact replicas. Similarly, many intangible creations can’t be effectively protected.
Thus, while deep neural networks can use transfer algorithms to assess stylistic similarities between works, even if two pieces share identical styles, the content might still be original and not constitute plagiarism.
(The Nude clothes hanger and its imitators)
Even for truly original works, the “qualitative standards” are hard to quantify. Infringers can tweak a few details and claim it as their own. While machine learning can spot subtle differences faster than the human eye, pursuing legal action against “derivative plagiarism” remains extremely difficult.
Shen Wenjiao’s Nude hanger, for instance, spawned countless modified versions on some websites, making it hard for official platforms to remove all knockoffs.
The famous novel Ghost Blows Out the Light: The City of Lost Souls saw its author, Tianxia Bachang, sue the production team of the derivative film Chronicles of the Ghostly Tribe for altering the original work. The case took four years to conclude, with a meager compensation of just 50,000 yuan.
The low return on investment for legal action forces many creators to resort to “calling out” infringers on social media, relying on moral condemnation to vent their frustration.
The hard truth is that while AI excels in many fields, it’s still immature when it comes to combating plagiarism and piracy, which involve “ideas.”
It sounds disheartening—does copyright protection really rely solely on conscience, with no effective solutions?
In a way, this issue is like a comprehensive final exam in modern civilization, unsolvable by a single technological formula. Yet, integrating new technologies is inevitable. In the meantime, building a “shield” around original content is an art of technological civilization.
Currently, some blockchain-based copyright platforms offer new hope for original content protection. Features like IPTM timestamps enable quick judgments during registration and infringement identification, with detection occurring during the upload process—processing thousands of images in as little as two hours, eliminating the “too little, too late” dilemma.
However, registering original works often requires uploading electronic certificates, author information, and content data to the blockchain, leading to significant computational and cost burdens that deter policymakers and businesses.
In the future, as deep learning algorithms optimize the allocation of distributed idle computing power, copyright platforms can operate more efficiently and affordably amid the explosion of digital content, raising the risks for infringers.
While AI struggles to discern “ideas,” it excels in improving judicial efficiency—directly addressing the core pain points of originality protection: difficult identification and lengthy, exhausting legal processes.
Today, smart judiciary systems can use machine assistance to comprehensively gather, analyze, and automatically compare material features. By handling these procedural and foundational tasks, judicial personnel can focus on complex aspects like hearings and deliberations, shortening case cycles and reducing维权 costs.
Machine learning can’t perceive “ideas,” and its ability to mimic human creations—whether images or text—remains rudimentary. But it can also help creators maximize the value and appeal of their ideas.
For example, Adobe’s suite of “must-have” tools for designers automates tasks like lighting adjustments and image cutouts.
Building 3D models in FE allows for one-minute renders. The less time spent on these basics, the sooner we’ll enter an era where human designers compete on creativity and originality, avoiding burnout.
As AI, cloud computing, blockchain, and other technologies mature, life is becoming increasingly intelligent. Machines, society, and humanity are forming a new “iron triangle” of future civilization.
This might be a new paradox: only through the evolution and support of machine civilization can the wisdom and brilliance of humanity shine so uniquely and deserve celebration.