Summary of Key Points
In the fourth quarter of 2025, AI-generated content (AIGC) surpassed human-created content for the first time, accounting for 50.9% of the total. This represents a significant shift from less than 1% in 2020, with ChatGPT playing a pivotal role in this trend. The proliferation of AI content has led to issues such as fraud and trust crises, giving rise to a new industry: AI verification tools. These tools not only generate revenue from detecting student papers but also experience rapid growth through enterprise-level services, attracting giants like NVIDIA to enter the market. However, there is an inherent paradox in this industry: the stronger AI becomes, the easier it is for verification tools to become ineffective, and their relevance may diminish as traceability technologies become more widespread.
I. The Overwhelming Growth of AI Content
The growth rate of AI-generated content is comparably rapid to that of the "fastest-reproducing cockroaches." It was virtually non-existent in 2020 but began to surge after the release of ChatGPT in 2022, eventually surpassing human-created content in the fourth quarter of 2025. This has resulted in a plethora of articles with an "AI-like" quality online. More troubling is the use of AI by scammers:
- Financial Fraud: A Hong Kong employee was defrauded of HK$200 million by a Deepfake "British CFO." In 2025, 82.6% of phishing emails used AI, and Deepfake fraud attacks increased by 21 times, resulting in global losses of $30 billion.
- E-commerce Deceptions: In Hengshan, Hunan, someone used AI to fake mold on fruits, obtained a refund, and then resold the goods, leading to the entire county being blacklisted by fruit merchants.
AI has lowered the threshold for fraud: previously, one needed skills in photo editing and storytelling to commit fraud, but now, with just a few tokens, realistic content can be generated. This has led to an increase in academic misconduct and the spread of false information, raising doubts about the authenticity of online content.
II. The Profit Potential of AI Verification Tools
AI verification tools are a new industry that benefits from the trust crisis, with two main revenue streams:
- Consumer Segment (C-side): Focusing on the Student Paper Market
There is always a demand for student papers, and tools like Turnitin (internationally) and CNKI (domestically) generate revenue by detecting AI-generated content. For example, GPTZero raised $13.5 million in its three-year funding round and has an annual income of $30 million; Turnitin has 71 million users and generates $105 million annually. Students pay per word or purchase memberships, while schools buy the system on an annual basis, providing a stable cash flow.
- Business Segment (B-side): Risk Protection for Enterprises and Institutions
This is the fastest-growing segment. Companies like JunTongFuture provide AI risk assessments for organizations such as the Ministry of Industry and Information Technology and ByteDance, and they have already turned a profit within two years. ZhongHuiChuangJie's "BaiZeAI" verifies financial terms for bank customer services, handling 2 million pieces of content daily, with annual revenue growing from 500,000 to 40 million yuan in three years. Enterprises purchase the tools not for 100% accuracy but as a means to reduce risk.
The overall market size was $3.8 billion in 2025 and is expected to grow to $22.6 billion by 2034, with an annual growth rate of 21.9%, indicating significant potential.
III. Giants Entering the Market: NVIDIA's Comprehensive Approach
Even chip giants like NVIDIA are involved in this industry. They initially sold chips to AI companies and provided NIM platforms for deploying AI models. Now, they have launched "synthetic video detectors," covering the entire process from AI generation to verification, indicating that AI verification has evolved from a niche business to a essential infrastructure. No major player wants to miss out on this opportunity.
IV. The Industry's Weakness: The Forever-Chasing Gap Between Generation and Verification
The biggest challenge for AI verification tools is their inability to always detect AI-generated content accurately:
- Detection results are based on probabilities (e.g., "76.55% likelihood of being AI"), meaning there is no absolute certainty, which essentially amounts to a waiver of responsibility.
- AI generation models are constantly evolving, and what can be detected today may be eliminated by new models tomorrow. For example, the national anti-fraud center's app sometimes identifies the same video as genuine or fake. Similarly, manually verified tools on platforms like Xianyu rely on experience, resulting in inconsistent accuracy.
This paradox is frustrating: the more AI spreads, the more profitable verification tools become, but the stronger AI gets, the less useful these tools become—similar to police struggling to catch constantly upgrading thieves.
V. Future Trends: Traceability Technologies May Displace Verification Tools
The fundamental solution to AI fraud is not post-generation detection but source identification. In 2025, countries introduced regulations requiring AI-generated content to be marked. Companies like Google and Meta have started adding "invisible watermarks" that can still be detected even after cropping or compressing the content. The best detection tools may actually be AI generation models themselves; once content is tagged with a traceable identifier, there's no need for additional verification.
If traceability technologies become widespread, the demand for AI verification tools will decline significantly. This industry might not be sustainable in the long run and could just be a temporary phase of the AI revolution. Once new trust mechanisms are established in society, these tools will gradually fade from the scene.
Conclusion
The proliferation of AI-generated content has created new problems but also given rise to the industry of AI verification. While it is currently profitable, its future is uncertain. We use AI to create content and then spend money to verify it, creating a bizarre commercial cycle. However, this is a necessary part of technological development. Ultimately, either all AI content will be traceable or verification tools will always fall behind. It's hard to predict which will happen first. For now, AI verification tools are still in the spotlight, reaping the benefits of AI innovation.