During the Spring Festival, DeepSeek became a sensation, almost rivaling the excitement surrounding the 2025 Spring Festival Gala. Initially, I dismissed it with a hint of disdain.

My skepticism stemmed from an image circulating online. A user had unsuccessfully requested pornographic websites from DeepSeek and then inquired, "I hope to block pornographic websites. Can you give me a list of sites that need to be blocked?" DeepSeek bluntly provided links to unhealthy websites...

At the time, my instinctive reaction was that DeepSeek was yet another product of opportunistic speculation and hype, much like the now-forgotten metaverse—flashy but ultimately useless.

However, after personally downloading and trying it, I was thoroughly impressed. DeepSeek's AI compares user questions with numerous webpage contents, understands them, and then filters and generates the desired answers.

In other words, DeepSeek's logic for blocking unhealthy websites is sound; it merely fell victim to clever human questioning. With AI's astonishing learning ability, searching the same question now yields normal results.

This raises an intriguing question: Will AI's role in internet content governance become even more extensive in the future? Will major search platforms finally eradicate internet "scourges" like pornography and vulgarity?

Consider Quark, a search engine favored by young users. A few years ago, Quark was still dubbed the second "Qvod" by many users due to the presence of pornographic and obscene information in its search results. Quark was officially fined and required to rectify the issue.

Even today, if you search for Quark on platforms like Xiaohongshu and Weibo, many comment sections still contain hints from so-called experienced users that you can find unspeakable video websites on Quark...

These pornographic contents that cannot be removed or eradicated have become the dark side of major search platforms, hidden from public view.

Quark is Just One AI Away from Shedding the "Qvod" Stigma

Setting aside other search platforms, Quark has the least history of involvement with pornography, judging from its past and present.

When the infamous Qvod collapsed around 2013 due to the "anti-pornography campaign," the battleground for disseminating unhealthy content shifted to PC-based cloud storage products subsequently launched by various internet companies.

According to Tianyancha APP, during this period, Ali also fully acquired UC (browser, cloud storage).

By 2016, the entire industry, led by Baidu Cloud Storage, encountered a "special rectification campaign" targeting the use of cloud storage to disseminate pornographic and obscene information. At that time, Ali's UC Cloud Storage even stopped its cloud storage service to cooperate with relevant departments in investigating pornographic and pirated content, and it was not relaunched until two years later.

It was precisely during this period of "severe crackdown" that Quark emerged from within Ali UC.

With such a precedent, how could Quark, as a late entrant, not learn from it?

Moreover, Quark once introduced itself as follows: "My biggest pursuit is to be a good smart search app. My ultimate goal is to maintain a pure search environment for all users, allowing those who possess me to also enjoy a focused experience free from advertisements and information streams." Isn't this like a pure and innocent little white flower? But who knew that the flower's core would gradually turn yellow over time.

At the end of 2023, the Cyberspace Administration of China imposed a fine of 500,000 yuan on Quark. The reason was that the "Quark" platform did not comply with relevant management requirements, presented a large amount of pornographic and obscene information in search results, and recommended pornographic and vulgar keywords to users...

In response, Quark said at the time that it "attached great importance and comprehensively rectified," but judging from the current situation, there was no reason for this mistake to occur in the first place, so why hasn't it been completely resolved until today?

There are multiple reasons: on a larger scale, this is not just a problem for Quark but a passive helplessness faced by the entire internet search industry when dealing with complex social issues related to human needs, which are indeed difficult to completely eradicate.

On a smaller scale, it may also be related to the scale of user acquisition and content review of searches.

For example, Quark initially highlighted its advantages of no advertisements and no push notifications, which were indeed attractive in terms of search experience, but for users, that was about it. After all, the core value of search lies in content output, such as knowledge, technology, and popularization. The lack of richness in these contents was a shortcoming of Alibaba and a relatively weak point for Quark in competition with search players like Baidu and Sogou (WeChat).

Under such circumstances, if Quark's content review is too strict, it will lack content appeal, and key data performance such as daily and monthly active users will suffer. Conversely, an overly lenient review scale can indeed achieve the dual goals of rich content and user acquisition, but it is easy to cross regulatory red lines.

Essentially, this is similar to how short video platforms like to push beauty videos and borderline content to male users. Although somewhat vulgar, within certain limits, it is indeed an effective way to attract users.

However, while borderline short videos may not be illegal, the search and dissemination of pornographic content are absolutely unacceptable.

In the past, when the industry was less regulated, Quark might have relied on unconventional methods and taken risks. But now, with Quark upgraded to the "Four Little Dragons of Alibaba" under Wu Yongming, and with Alibaba's "Tmall Genie," one of its few ToC hardware brands, expected to collaborate with the "Quark" team to explore new hardware directions, including AI glasses, Quark, which integrates software and hardware, has clearly been entrusted with the hope of becoming the main front for Alibaba's future AI-to-C strategy.

Therefore, Quark must now completely eradicate the last vestiges of the internet "scourge" left by past content violations.

In this regard, the emergence of DeepSeek provides a breakthrough idea for search platforms like Quark from the perspective of AI regulatory review, playing a more proactive role in blocking unwanted content.

But that's not enough. If Quark cannot find a more effective user acquisition path, it will be difficult to truly tighten the content review gap.

Fortunately, Quark has now found the answer, relying on AI or rather AI tools.

Realizing that it was difficult to surpass Baidu in content supply, Quark search began emphasizing its innovation at the "AI tool" level very early on, launching anti-epidemic AI tools like "Fellow Traveler Inquiry" and "Doctor Online" during the pandemic. Now, when opening the Quark App, the first application tools visible include cloud storage, Scan King, Learning, Daily Report, Documents, and Hot Search. In the AI tool pack, there are also AI PPT, AI resume, etc., all focusing on services for practical application scenarios.

Subsequently, the strategy of combining search content, AI tools, and service experience indeed became Quark's breakthrough point for user acquisition. For example, in the summer of 2021, the advertising slogan "Choose the right volunteer, just use Quark" instantly made the "Quark College Entrance Examination" AI search a top choice for countless parents and students after the end of the college entrance examination season each year.

Even until now, in the report "Ten Trends in the Development of the AI Industry in 2025" released by Analysys, Quark ranks first on the list of active users of AI applications in China in 2024, with a MAU of 71.02 million. It is also the only AI application where the number of post-00s young users accounts for more than 50%.

Against this background, with Alibaba's attention and Deepseek's explosion, whether Quark wants to or not, as the "AI all-rounder assistant for 200 million people," it must personally tear off the past Qvod label and use AI to wash itself clean and prove itself...

How Does Quark Address the Commercial Value of AI Search?

After discussing the assistance of AI to Quark and others, let's turn to the contradiction between large models and Quark search.

When results can be obtained with just a few words, who would still want to use a search engine to find information? When answers can be obtained directly, who would still want to click through webpage links one by one and struggle to choose between real hospitals and Putian-affiliated hospitals, real official websites, and fake advertisements?

Therefore, there is bound to be a contradictory and substitutive relationship between large models and AI search.

However, current large models have not yet evolved to the point where they can replace search engines. To give an intuitive example, when I ask "How to fix a dripping faucet" in Tongyi APP and DeepSeek, all the responses are textual descriptions, and relevant images cannot be provided directly. But the Quark APP is different; it offers textual answers, step-by-step images, and a bunch of similar repair videos for reference.

So who can best solve my problem? It's obvious.

Therefore, considering future and current commercialization, in the face of the uncertainty of large models and the nearly certain growth of AI search, it is not surprising that Alibaba, ByteDance, and others are now increasing investment in both large models and AI search.

The only thing to note is that growth is the path to value, not value itself.

Especially after experiencing the bubble of trends like the metaverse, both large models and the already relatively mature AI search now need to prove their commercialization potential to the outside world.

So, against this background, how is the money-making ability of AI search applications led by Quark?

It may be somewhat unsatisfactory.

There are two reasons: one is the limited advertising value. Nowadays, many AI searches have chosen a positioning of few or no advertisements to attract users. For example, ByteDance's Wukong Search and Baidu's "Baidu AI Search" follow this approach.

As for Quark, it goes without saying that its initial product positioning was ad-free and precise, which successfully carved out a niche in the search engine market dominated by Baidu at the time and became the reason why many young people have always "supported" it.

So, what is Quark's first label? The minimalist style of no advertisements!

But this is undoubtedly contrary to the main (advertising) commercialization strategy of search engines. Moreover, because Quark initially established a brand persona of "few advertisements," even if Quark now wants to add some "earning" advertisements, it is easy to cause dissatisfaction among its core user base...

The second reason is the immaturity of AI tool payments. Since the content ecosystem cannot compete and the advertising route is not viable, Quark and others can only choose to be a versatile "super APP" toolbox and move towards user membership subscription services.

The problem is that Quark does indeed have many functions, but there are also many corresponding membership categories. There are five independently charged membership systems within a single Quark App: Cloud Storage, Scan King, Documents, Learning, and Novels.

According to previous actual comparisons, under the new user limited-time discount, the continuous annual subscription price for Quark Cloud Storage SVIP is 158 yuan, which is even higher than Baidu Cloud Storage's 154 yuan.

It should be noted that the product logic of cloud storage is somewhat similar to the network effect of social software; the more people who use it, the more people will use it. Therefore, there are few cases in the internet industry where small players can win. It is always the big and strong players who dominate the market.

With Baidu Cloud Storage once accounting for over 80% of the domestic personal cloud storage market, as a latecomer, it is not easy for Quark to grab more market share. In this context, will the current high membership pricing be an additional burden?

Therefore, even though Alibaba has not disclosed Quark's performance in its financial reports, it seems that Quark's own commercialization journey may not be smooth. As for the rumored integration with the Tmall Genie team to jointly develop AI glasses and other smart hardware products, it is still far from certain and completely unknown.

But if this is the case, if Quark cannot demonstrate long-term strategic value or seek more commercialization potential in the short term, then even if Quark is currently highly valued, its future position within Alibaba will definitely be precarious.

Even the new retail business that was favored by Jack Ma has now been scaled back as Alibaba focuses on core businesses due to prolonged deficits exceeding revenues. The landmark event was the successive sales of Intime Department Store and RT-Mart.

As for the remaining Hema, it has also begun to tighten its belt "for survival" and strive to achieve profitability...

Taking this as a lesson, how many years of "trial and error" will Alibaba allow Quark for its commercialization closed loop in the future?

After all, the story of changing rulers repeats itself every day in the AI and internet industries...