In December 2025, ByteDance's "Doubao" large-scale model unveiled a system-level mobile assistant and forged a partnership with Nubia M153, drawing considerable attention.

The Double-Edged Sword Effect Becomes Prominent

This product, which claims to "operate smartphones like humans," centers on cross-app intelligent scheduling and natural language interaction as its core innovations, aiming to disrupt the traditional touch-based operation paradigm. Upon launch, it attracted significant market interest.

The initial batch of 30,000 engineering prototypes, priced at 3,499 yuan, sold out rapidly. On secondary markets, resale prices soared to nearly 10,000 yuan, with priority reservation codes fetching 300 yuan and daily rental rates reaching 500 yuan, mirroring the market fervor seen with early disruptive tech products.

However, this enthusiasm is accompanied by escalating controversy. The product has been mired in rumors of "regulatory inquiries" due to cybersecurity and data privacy concerns stemming from its system-level access. Although on December 13, informed sources clarified to The Paper that "the news is false," discussions about its security have not abated.

In reality, ByteDance has explicitly ruled out plans for developing its own smartphones. The Doubao Mobile Assistant is a software product built on deep collaboration between the app and manufacturers at the operating system level. Activated via voice commands or side AI buttons, it can manage complex tasks such as "cross-platform price comparison and ordering" or "organizing travel plans and syncing them to WeChat." It requires manual confirmation only for sensitive actions like payments while prohibiting risky operations such as score manipulation or financial app usage.

Nevertheless, real-world testing revealed that the assistant took nearly two minutes to compare milk tea prices across three food delivery platforms, significantly slower than manual operations, and erroneously included Taobao proxy-ordered items in price comparisons.

Some users attempting WeChat transfers triggered the platform's security controls, resulting in temporary account restrictions. These technical flaws and security vulnerabilities have further underscored the dual nature of system-level permissions.

The core value of the Doubao Mobile Assistant lies in its fundamental shift from "tool empowerment" to "intelligent agency," offering possibilities to reshape the mobile internet ecosystem.

From a user perspective, it breaks down traditional app silos, freeing users from cumbersome multi-app switching. Whether it's one-click cross-platform price comparison and coupon collection after being inspired by content or automatically organizing travel plans on Xiaohongshu and seeking friends' opinions, it significantly lowers the barriers to digital life. Especially in scenarios like driving or working, voice interactions enhance efficiency, elevating the mobile assistant from a passive tool to an active "digital life assistant."

From an industry standpoint, it provides a new direction for the stagnant smartphone industry, breaking free from the homogenization of hardware competition and driving the transition from "hardware terminals" to "AI service gateways."

For ByteDance, this is a critical move to secure an ecological foothold in the AI era. Through a lightweight "large model vendor + smartphone manufacturer" collaboration, it can become the core intermediary between users and digital services without developing its own hardware, potentially disrupting the current traffic dominance of super apps. For traditional hardware makers like ZTE, it represents a vital attempt to differentiate through AI.

Building Exclusive Ecological Barriers

The emergence of the Doubao Mobile Assistant signals the dawn of the AI agent era, reflecting clear industry trends while concealing multiple risks.

At the trend level, deep collaborations between "large model vendors and smartphone manufacturers" will become the mainstream paradigm. Google's partnership with Samsung and Doubao's collaboration with ZTE, along with ongoing negotiations, further validate this complementary and win-win approach.

The smartphone industry will shift entirely toward "AI service capability competition," with consumer demands evolving from "high performance" to "intelligent services." AI assistant interaction fluidity and task efficiency will become core competitive factors, potentially reducing hardware profit margins while making value-added services a new growth driver.

The mobile internet ecosystem will undergo a "de-appification" transformation, with AI agents emerging as new service gateways, weakening the centralized role of app stores and promoting decentralized service distribution.

AI application compliance regulations will continue to tighten. The EU and the U.S. have already enacted relevant laws, and China is advancing AI-generated content traceability, with permission boundaries and data norms becoming regulatory priorities.

Risks cannot be overlooked either.

Technologically, large models exhibit uncertainties. Beyond efficiency shortcomings and recognition flaws observed in testing, instruction misinterpretations could lead to errors like "booking a business-class flight instead of an economy one." In sensitive scenarios involving payments or personal information management, this could result in financial losses or privacy breaches. Officials also caution that "demo scenarios may not be fully replicable."

In terms of industry competition, tech giants and platforms have launched ecological defense strategies. Google swiftly upgraded the Gemini assistant's permissions on Pixel phones, enabling cross-app schedule management and email sorting, directly competing with Doubao's features.

Apple leverages its closed iOS ecosystem to restrict third-party AI assistants from accessing system-level interfaces, granting only basic permissions while accelerating its own on-device large model development, planning to upgrade Siri into a cross-app intelligent agent.

Huawei has deeply integrated its Pangu large model with the HarmonyOS, launching "AI-native services" that enable multi-terminal collaboration across phones, watches, and cars, constructing exclusive ecological barriers.

This has triggered ecological rivalries.

Traditional super apps have launched more direct countermeasures. Some banking apps now mandate disabling AI assistants for payment functions, while WeChat and Alipay have tightened risk control algorithms, strictly limiting AI-proxy transactions and transfers, even triggering temporary account freezes. Platforms like Meituan and Douyin have introduced "non-human operation detection" mechanisms in the background, throttling AI-driven bulk price comparisons and automatic likes, fueling ecological rivalries.

Regulatory scrutiny on permission management and data protection is intensifying, with clear policy signals. The EU's AI Act classifies AI assistants with system-level permissions as "high-risk AI systems," requiring compliance obligations like data logging, risk assessments, and human oversight.

China's Interim Measures for the Management of Generative Artificial Intelligence Services also emphasize algorithm registration and data security reviews. Recently, the Ministry of Industry and Information Technology launched a "special AI terminal device security rectification," focusing on investigating permission abuses and data leaks in system-level AI assistants.

Legal issues such as service liability definitions remain unresolved. If disputes arise from transactions completed via AI assistants, responsibility attribution is unclear, posing higher compliance demands on enterprises.

To address this, companies can adopt three core compliance strategies. First, establish a "tiered permission management" system, categorizing user data by sensitivity and granting AI assistants only the minimum permissions necessary for task completion.

Second, implement "operation traceability" requirements, logging every cross-app action by AI for regulatory review and liability determination. Introduce "human oversight checkpoints," mandating manual confirmation for payments and identity verification to reduce risks.

In terms of user perception, survey data reveals significant stratification in market acceptance. According to a Q4 2025 smartphone user survey by a third-party agency, 72% of users aged 18–30 embrace AI mobile assistants, with over half willing to pay for premium features. However, only 28% of users over 40 accept them, with 62% citing "privacy concerns" and "distrust in AI accuracy." Additionally, 35% of middle-aged and elderly users frankly state they "feel more at ease with manual operations."

Follow-up interviews with early adopters show that 30% abandoned use due to "operation errors causing trouble," while 25% uninstalled it over "fears of losing basic skills through over-reliance on AI." These figures underscore how user cognitive biases constrain product penetration and highlight long-term challenges for enterprises in user education and product optimization.

Ultimately, only by balancing technological innovation with compliant development and fostering ecological collaboration can AI mobile assistants evolve from concept to maturity, truly achieving efficient and intelligent upgrades in the digital ecosystem.