The Dispute Between Doubao and Major Tech Platforms: Legal Challenges and Compliance Analysis of AI Smartphones

Introduction: A Systemic Conflict Triggered by "Proxy Operations"

Recently, a seemingly modest user experience has sparked heightened tensions between the AI industry and internet platforms. Certain smartphones equipped with AI assistants, when attempting to automatically complete tasks such as sending WeChat red packets or placing e-commerce orders via voice commands, were identified by platform systems as "suspected use of unauthorized plug-ins," triggering risk warnings or even account restrictions.

On the surface, this appears to be merely a technical compatibility issue; however, within the broader industrial context, it essentially unveils a structural conflict surrounding"who has the authority to operate the smartphone and who controls the user interface."

On one side are smartphone manufacturers and large language model teams seeking to deeply embed AI into operating systems to achieve "seamless interaction"; on the other are internet platforms that have long relied on app interfaces, user pathways, and closed data loops to build their commercial ecosystems.

As the "universal assistant" begins to act on behalf of users, is it an efficiency tool or a rule-breaker? This question is now being brought before the law by reality.

 

"The Future Is Here" or "Risk Warning"—A "Code War" Behind the Smartphone Screen

Recently, users who obtained the latest AI smartphones may have experienced a dramatic scenario of "one second in the future, one second under warning": just as they marveled at the convenience, they received risk notifications from platforms such as WeChat.

This all began with the deep collaboration between ByteDance’s "Doubao" large language model and certain smartphone manufacturers. Today’s voice assistants are no longer limited to checking the weather; they have become super-butlers capable of "seeing the screen and simulating operations."

Imagine the following scenario:Simply say to your phone, "Send a red packet in the Qingfei Football Team group chat" or "Help me buy the most cost-effective new Adidas football shoes," and the phone will automatically navigate to the app, compare prices, and complete the payment—all without any manual intervention on your part.

This technology, based on "simulated clicks" and "screen semantic understanding," has for the first time enabled AI to truly take control of mobile phones. However, this "seamless" experience quickly encountered firm resistance from internet platforms.

Many users have found that using Doubao AI to operate WeChat triggers account restrictions and even warnings of "suspected use of unauthorized third-party plugins." E-commerce platforms such as Taobao are also highly vigilant against such automated access. One blogger likened AI to a butler running errands on your behalf, only to be stopped by mall security: "We do not serve robots."

  • Users are puzzled: Why can't my AI, authorized by me and running on my own phone, act on my behalf?
  • Platforms maintain: Our ecosystem and our security do not permit external "proxy operations."

What appears to be a minor friction over technical compatibility is in fact another milestone confrontation in the history of China's internet. It is no longer a simple battle for traffic, but a direct clash between operating systems (OS) and super apps over "digital sovereignty."

 

A Dimensional Strike in Business Logic—When "Walled Gardens" Encounter "Wall-Breakers"

Why have major tech companies such as Tencent and Alibaba reacted so strongly? The answer lies in the core business model of the mobile internet: the "walled garden."

The commercial foundation of social media, e-commerce, and content platforms rests on exclusive entry points and user engagement time. Every click and every browsing step is key to advertising monetization and data accumulation. The emergence of "system-level AI assistants" like Doubao poses a direct challenge to this model.

This is a profound contest over "entry points" and "data." AI-powered phones have touched the core commercial lifelines of internet giants, primarily in three respects:

1. The "No Need to Tap App Icons" Crisis:

When users can simply issue voice commands and have AI complete tasks directly, the apps themselves may be bypassed. Users no longer need to open apps to browse products or view ads, which means the advertising exposure and attention economy on which platforms depend will be significantly weakened.

2. "Parasitic" Acquisition of Data Assets:

AI operates and reads information by "viewing" the screen, without requiring the platform to open an API. This effectively bypasses traditional cooperation rules and directly accesses content, products, and data that the platform has invested heavily in building. From the platform's perspective, this constitutes "free-riding," and there is even a risk that such data may be used to train the AI models themselves.

3. Change of Guard in Traffic Distribution Gatekeeping:

In the past, the power to distribute traffic rested with super apps. Today, system-level AI is becoming the new "master switch." When users ask for recommendations, the AI's response will directly determine where commercial traffic flows, which is sufficient to reshape the competitive landscape.

Therefore, the warnings and protective measures issued by platforms are not merely technical exclusion, but fundamental defenses of their own business ecosystems. This reveals deep-seated, unresolved tensions between technological innovation and platform rules.

 

Preparations Before the Storm—An In-Depth Analysis of the Fourfold Legal Risks of AI Phones

As legal practitioners, looking through the lens of the dispute between AI phones and major tech companies, we can identify four unavoidable core legal risks:

I. Competitive Boundaries: Technological Neutrality Does Not Equate to Liability-Free Intervention

The current focus of the controversy is whether AI operations constitute unfair competition. Under the Anti-Unfair Competition Law, using technical means to impede the normal services of another party's network products may constitute infringement.

  • Risks Associated with "Plug-ins":In the case of Tencent v. 360 and numerous recent cases involving automated red-packet grabbing plug-ins, judicial practice has established a principle: unauthorized modification or interference with the operational logic of other software, or increasing server load through automated means, may constitute unfair competition. If AI's "simulated clicks" skip advertisements or bypass interaction verifications, thereby affecting platform services or business logic, they may similarly face findings of infringement.

  • Traffic Diversion and Compatibility Issues:If AI guides users away from the original platform to use its recommended services, it may involve "traffic hijacking." Conversely, if a platform imposes a blanket ban on all AI operations, it may need to demonstrate that such bans constitute reasonable and necessary self-protection.

II. Data Security: Screen Information Constitutes Sensitive Personal Information

AI needs to "see" screen content to execute commands, which directly implicates the strict regulations under the Personal Information Protection Law.

  • Processing of Sensitive Information: Screen content often contains sensitive personal information such as chat records, account details, and location trajectories. In accordance with the law, obtaining the user's "separate consent" is required. The validity of the "bundled authorization" commonly seen in current AI-enabled smartphones remains questionable. If the AI, while executing a ticket-booking command, "sees" and processes private chat messages, it may violate the principle of "minimum necessity."
  • Ambiguity Regarding Responsible Parties: Does data processing occur locally on the smartphone or in the cloud? In the event of a data breach, how should liability be allocated between the smartphone manufacturer and the AI service provider? Current user agreements often fail to clearly define these responsibilities, creating compliance risks.

III. Antitrust Controversy: Do Platforms Have the Right to Refuse AI Access?

Future litigation may center on the concepts of "essential facilities" and "refusal to deal."

  • Proponents of AI-enabled smartphones may argue that platforms such as WeChat and Taobao have acquired the attributes of public infrastructure. Unjustified refusal to allow AI access may constitute an abuse of market dominance and hinder technological innovation.
  • Platforms may contend that data openness must be premised on security and the protection of property rights. Allowing AI to access data without authorization may circumvent technical protection measures, thereby harming the interests of both users and the platform.

IV. User Liability: Who Bears the Cost When AI Makes Mistakes?

As AI evolves from a tool into an "agent," it raises a series of issues concerning civil liability.

  • Validity of Agency Acts:If an AI purchases the wrong product due to misinterpretation (e.g., interpreting "affordable phone" as a counterfeit device), does this constitute a significant misunderstanding or improper agency? Can the user claim a refund on the grounds that the transaction was "not performed by the user personally"?
  • Losses from Account Bans:Where a user’s use of AI features results in the banning of a third-party account, the user may seek compensation from the mobile device manufacturer. The key issue is whether such risks were clearly disclosed at the point of sale. If disclosures are inadequate, the manufacturer may face collective claims by affected users.

This contest is not merely a technological dispute; it also involves redefining, in practice, the legal boundaries of data property rights, platform liability, and user authorization. Both AI vendors and platforms must identify a clear balance between innovation and compliance.

 

Conclusion: Boundaries of Rights and the Spirit of Contract

The friction between Doubao and major technology companies appears on the surface to be a product conflict, but in reality it reveals fault lines between old and new orders: the past app-centric model is encountering disruption from an AI-driven interconnected experience.

As legal practitioners, we clearly observe that the existing legal framework appears ill-equipped when confronted with the integration of general-purpose artificial intelligence. Mere “bans” or “workarounds” cannot constitute sustainable solutions. The way forward may lie not in continued reliance on technical circumvention such as “simulated clicks,” but in promoting the establishment of standardized AI interaction interface protocols.

At a time when the rules remain unclear, we pay tribute to those who persist in exploring the AI frontier and uphold the principle that technology should serve the public good. At the same time, we must soberly recognize that:Respect for boundaries often enables more enduring progress than disruption itself.