Interface design has undergone a gradual evolution from function-oriented to experience-oriented approaches, and further toward algorithm-driven and behavior-guiding models. Early Human-Computer Interaction (HCI) design emphasized improving operational efficiency, focusing on clear information architecture and intuitive operation. With the advent of the Web 2.0 era, interface design shifted toward prioritizing user experience and emotional interaction, gradually developing diverse design strategies such as visual guidance, process-oriented interaction, and personalized adaptation. In recent years, driven by data-centered design and context-aware computing, the interface has come to be regarded not only as a technical presentation layer, but also as a medium for behavioral guidance, emotional modulation, and commercial conversion.

Dark Patterns are a particular type of interface design formed under algorithmic decision-making and behavior-guidance mechanisms. Their purpose is to influence user choices through default options, visual cues, nested interaction pathways, and structural arrangements. The term Dark Patterns was first introduced by user experience designer Harry Brignull in 2010 to describe interface designs that deliberately mislead users into making decisions that benefit the platform while being detrimental to their own interests, often under conditions of incomplete awareness or limited understanding [8]. It is precisely due to this characteristic of “appearing neutral while subtly guiding” that Dark Patterns have increasingly become a highly controversial interface design strategy within digital platforms [8].

In contrast, “persuasive design” or “nudging” techniques aim to guide users toward beneficial behaviors while preserving freedom of choice. Dark patterns, by contrast, prioritize platform interests through deception and constraint, thus representing an ethically problematic form of manipulation.

As a “mediating layer” in human-computer interaction, Dark Patterns exhibit several key characteristics, including technical dominance, inducement, control, and concealment.

First, technical dominance. Designers of Dark Patterns hold overwhelming advantages in interface construction and interaction, user data collection and utilization, algorithmic interpretation, and continuous optimization, which constitutes a form of technological monopoly that allows platforms to guide, induce, or even manipulate users covertly [9]. As Brignull notes, the essence of Dark Patterns lies in leveraging users’ cognitive blind spots, habitual operations, and contextual dependencies to construct an asymmetric choice environment, so that users complete platform-preferred actions under an illusion of freedom [8].

Second, inducement. Dark Patterns often deploy visual or psychological strategies—such as highlighting recommended options, encouraging language, loss-framed prompts, countdown timers, and scarcity emphasis—to steer users toward platform-benefiting behaviors in moments of reduced deliberation; for instance, e-commerce promotions frequently present prompts like “only a few items left” or “most users have already purchased” to manufacture urgency and scarcity, thereby accelerating decisions [3].

Third, control. Although users appear to retain freedom of choice, their decisions are effectively made within a restricted and manipulated cognitive environment; this systematic shaping and guidance of behavior amount to a subtle yet highly efficient form of technological domination [7]. In practice, attempts to cancel subscriptions, log out, or disable features often encounter multi-layered menus and repeated confirmations; for example, video platforms commonly pre-select automatic renewal, while disabling it requires navigating several hidden interface layers [6].

Finally, concealment. Dark Patterns operate as a kind of covert discipline: through ambiguous terminology, procedural complexity, hidden pathways, and visual prioritization, platforms present an appearance of “free choice” while steering users away from fully rational decisions; for example, key fee items in loan application flows may be buried under “expand more” menus or visually downplayed to reduce salience [4].

Different types and intensities of Dark Patterns vary across interface environments [10]. Based on their influencing mechanisms and degrees of manipulation, Dark Patterns can generally be divided into inductive, deceptive, and coercive types [7].

First, inductive Dark Patterns do not directly restrict user choices, but instead guide users toward platform-preferred outcomes by setting default options, highlighting recommended items, or diminishing the visibility of alternative choices. The core mechanism here is the default effect, meaning that when a particular option is pre-selected, users are more likely to accept it rather than actively modify it [2]. This tendency becomes especially pronounced in situations involving high cognitive load or decision uncertainty.

Second, deceptive Dark Patterns mislead users by means of information concealment, lengthy text, vague terminology, and linguistic framing, causing them to make platform-favorable decisions without fully understanding the implications. This type of manipulation primarily exploits cognitive biases and information asymmetry. For example, in a study of 26 Norwegian financial applications, Raković and Inal found disguised pricing structures, hidden fees, and complex cancellation pathways to be widespread—particularly in non-bank financial products [11].

Third, coercive Dark Patterns influence user decision-making through forced procedural steps, time pressure, and emotional manipulation, thereby eroding users’ substantive freedom of choice. The core mechanisms involved include loss aversion, time urgency, and psychological pressure. For instance, in some subscription services, users attempting to cancel are required to navigate multi-layered menus or complete lengthy confirmation procedures, and may even be forced to call customer service or fill out detailed surveys, significantly increasing the difficulty and psychological burden of cancellation. This structure resembles a form of soft coercion [12].

In practice, however, many platforms use these forms in combination: beginning with inductive design to lower user vigilance, followed by deceptive wording to obscure key information, and finally applying coercive mechanisms to ensure alignment with platform-preferred outcomes. This progressive architecture of influence has become a core behavioral control strategy in contemporary digital platform environments.

The effectiveness of Dark Patterns in influencing user decisions lies in their strategic use of multiple design techniques during the interface structuring and interaction process, including default settings, textual terminology, path guidance, and structural arrangement.

First, default settings. Digital platforms frequently set options favorable to data collection, additional features, or subscription renewals as the default state. If users do not actively change these settings, consent is assumed. In practice, users tend to maintain default configurations rather than expend cognitive effort to locate and modify them—this phenomenon is known as the “default effect” [6]. Due to users’ limited understanding of data usage consequences, they often remain unaware of the behavioral costs or privacy risks embedded within default options.

Second, textual terminology. In privacy policies, data authorization statements, and responsibility clauses, platforms often employ ambiguous wording and lengthy descriptions, which obscure the actual meaning and consequences of user selections. As Miranda Fricker notes, professionalized language can effectively exclude many users from meaningful cognitive participation, preventing them from fully understanding their situational context [13]. Even when users complete the required actions of “reading” and “agreeing,” they may still not comprehend the real implications of their decisions.

Third, path guidance. Beyond linguistic framing, digital platforms rely heavily on non-verbal interface elements—such as color contrast, button size, visual focus, and information hierarchy—to direct users’ attention and behavioral patterns. These subtle guidance mechanisms increase the likelihood of users selecting platform-preferred options, while ignoring alternatives that may better protect their interests. Gray et al. argue that the essence of such “cognitive disciplining” lies in which information is made most visible [4]. By following visually suggested routes, users often complete intended operations without reflection.

Finally, structural design. Platforms strategically manipulate interface structure so that options beneficial to users become inconvenient, hidden, or difficult to access, while options advantageous to the platform are made convenient, salient, and easily accessible. For instance, cancellation pathways may be buried under multiple menu layers, while account logout may be intentionally complicated [12]. Moreover, Dark Pattern design often forms a “trigger–feedback–dependency” loop, cultivating behavioral habits that invisibly guide and reinforce user choices [14].

Dark Pattern design is not an incidental interface flaw, but a deliberately engineered apparatus—a comprehensive technological manipulation system jointly driven by big data, algorithmic inference, behavioral science, and interface aesthetics.

First, Dark Pattern design is grounded in the deep extraction of user information. Platforms do not merely collect users’ basic identity information and browsing behaviors; they also analyze fine-grained behavioral traces such as click pathways, time spent on specific elements, and preference signals. These micro-behavioral indicators are used to construct accurate user profiles and predictive behavior models. Through continuous updates enabled by deep learning methods, platforms are capable of inferring user intentions with increasing precision, thereby formulating customized interface manipulation strategies. For example, a platform may analyze a user’s response latency and frequency when interacting with “exit” or “decline” buttons in past interfaces, infer the user’s psychological threshold or hesitation pattern, and accordingly adjust the intensity of persuasive prompts or the visibility of exit pathways.

Second, Dark Pattern design relies on meticulously structured interfaces as the operational medium. Interface design typically includes information architecture, visual guidance, linguistic and symbolic cues, and interaction feedback. Under ordinary circumstances, these techniques help reduce cognitive load and improve user experience. However, in Dark Pattern design, these elements are strategically repurposed as tools of control. In information architecture, platforms may increase the number of layers required to cancel a service; in visual presentation, acceptance options are highlighted using strong color contrast while rejection options are visually minimized; in linguistic expression, ambiguous or euphemistic phrasing is used to conceal material consequences; and in interaction feedback, friction and psychological pressure are deliberately crafted to prevent user actions that may disadvantage the platform [15]. As scholars have noted, Dark Patterns in privacy interfaces are not isolated design accidents, but represent a systematized mechanism of interface manipulation [3].

Third, Dark Pattern design is fundamentally enabled by multiple algorithmic techniques. The functioning of the interface depends on user profiling, algorithmic prediction, and behavioral modeling. User profiling enables platforms to identify preference structures and habitual tendencies, which directly inform default-setting strategies, pop-up timing, and interface prompts. Algorithmic prediction is used to detect the moments when users are most susceptible to influence, allowing platforms to precisely deliver “exclusive offers,” “time-limited deals,” or “scarcity messages.” Meanwhile, behavioral models, drawing on psychology and behavioral economics, embed cognitive principles such as loss aversion, time pressure, and confirmation bias into interface interactions, thereby shaping user decision-making processes [16].

The effective functioning of Dark Patterns relies on a deep understanding of human cognitive patterns and psychological mechanisms, enabling users to perform platform-preferred actions while believing they are acting autonomously.

First, dark pattern design exploits users’ knowledge limitations. There is a significant cognitive gap between technology designers and ordinary users in terms of algorithmic literacy, interface logic, and terminology comprehension. Most users lack the ability to recognize the design intentions and potential risks embedded in interface language, making it difficult for them to accurately evaluate what they are being guided to do [17]. Research by Eslami and colleagues shows that users are often unaware of the presence of algorithms and misinterpret recommendation results based on everyday assumptions rather than actual computational mechanisms [18].

Second, dark pattern design takes advantage of informational asymmetry. Interfaces frequently employ information masking strategies—technical jargon, obscure phrasing, lengthy clauses, and complex procedures—so that information is formally disclosed but substantively incomprehensible, producing a condition of “apparent transparency but practical opacity” [19]. Under limited rationality and decision fatigue, individuals tend to rely on heuristics, which allows platforms to influence decisions by manipulating default options, persuasive wording, and visual emphasis [2].

Third, dark pattern design exploits users’ psychological limitations. Users are not always rational decision-makers. Platforms deliberately create a sense of urgency, scarcity, or social conformity, prompting immediate emotional responses instead of reflective judgment. Recent behavioral economics and HCI research has shown that dark pattern strategies are frequently built upon psychological mechanisms such as the default effect, visual attention bias, and loss aversion [7][20]. For instance, during account registration on social media or e-commerce applications, users are often required to agree to privacy policies before proceeding. These documents are typically long, technical, and filled with ambiguous legal terminology, making it difficult for users to fully understand the implications and risks in a short period of time.

In e-commerce platforms, dark patterns are commonly embedded in key processes such as product recommendation, payment checkout, and after-sales service. Their primary function is to increase purchase conversions and maximize platform revenue. For example, checkout pages often pre-select options such as automatic renewal or bundled add-on purchases, causing users to incur additional charges if they fail to manually deselect them. Furthermore, platforms frequently display time-limited or scarcity-based prompts to induce quick, emotionally driven decisions.

In cancellation or refund processes, users are often required to navigate multiple interface layers, repeated confirmation steps, or even customer service communication—creating path obstacles that discourage or prevent withdrawal. A well-known case is Amazon Prime, where cancellation requires navigating multiple redundant pages and repeated confirmation steps, representing a typical path-obstruction dark pattern [22]. According to the 2023 report of the European Insurance and Occupational Pensions Authority (EIOPA), such interface designs significantly increase the uptake of additional insurance products but simultaneously compromise user rights and information transparency, thus constituting clear instances of dark pattern manipulation [23].

As dark pattern design extends from commercial platforms to public service sectors, it increasingly shapes user decisions in ways that favor institutional interests.

In the healthcare sector, dark patterns are often embedded in processes such as patient data authorization, informed consent, and subscription to medical services. Studies have shown that many digital health platforms—for example, telemedicine applications and chronic disease management systems—pre-select the option “data used for research purposes” during initial registration, while burying the cancellation path behind multiple layers of settings [24]. Meanwhile, clinical trial recruitment platforms frequently employ ambiguous terminology and fragmented disclosure, placing key risk-related information in non-prominent interface positions. For example, some targeted cancer therapy trial interfaces use phrases like “data will be used to improve future health outcomes” instead of explicitly stating that such data may be sold for commercial purposes, thereby framing consent in a more ethically appealing manner. Such practices pose significant challenges to the ethical foundations of public healthcare [25].

In the education sector, dark patterns are quietly influencing how individuals access knowledge and educational services. Online learning platforms in particular routinely employ strategies such as “default subscription,” “automatic course renewal,” and emotionally suggestive prompts to retain users. Research indicates that platforms such as Coursera and Udemy often pre-select automatic renewal and periodic notification functions at the time of registration, while the paths to disable these features are hidden within multiple layers of settings. During the cancellation process, users are frequently confronted with prompts such as “You may miss valuable learning opportunities” or “Your skill improvement window is closing,” leveraging the psychology of loss aversion to reinforce platform control [26].

In legal consultation and job recruitment services, dark pattern design is often reflected in the complexification of contractual texts. Long passages of professional terminology and ambiguous expressions are frequently used to obscure key clauses, leading users to accept additional fees or restrictive conditions without fully understanding their implications. For example, payment obligations, liability statements, and risk clauses are often embedded deep within lengthy agreements, prompting users to click “agree” out of convenience while overlooking potential legal consequences. As King and Stephan note, such interface strategies appear to uphold the principle of informed consent, yet rely on “structural legitimacy” in interface design to circumvent substantive disclosure obligations, thereby maximizing the platform’s interests [27].

In social and dating applications, dark pattern design is more closely associated with the manipulation of attention and emotional engagement. Social platforms curate the rhythm of interactions to encourage continuous participation, response, and feedback, fostering a sense of dependency on the platform. In their analysis of platforms such as Facebook, Instagram, and Tinder, Gray and colleagues highlight the widespread use of “interface hooks”—design elements that intentionally guide user attention and behavior to extend engagement time and increase data production [4]. For example, some popular dating apps hide the “unmatch” or “remove connection” buttons within multiple layers of menus, or display prompts such as “Are you sure you want to give up a potential connection?” before allowing disconnection, using emotional framing to reinforce user retention. Likewise, in privacy settings, options such as “display profile to strangers” or “sync contact list” are frequently pre-selected by default. If users do not actively opt out, they may unintentionally disclose personal information [28].

As dark patterns become increasingly prevalent, regulatory and legislative bodies around the world have begun to address the impacts of such practices on user privacy, data security, and consumer rights. The European Union’s General Data Protection Regulation (GDPR) established one of the world’s most stringent data protection systems, requiring platforms to provide transparent consent processes and prohibiting the collection of user data through implicit consent or pre-selected options. The regulation further stipulates that users must be clearly informed of the meaning and consequences of their choices, and must retain the right to withdraw consent [31]. In contrast, privacy protection in the United States is more decentralized, relying on state-level legislation and sector-based norms. Among them, the California Consumer Privacy Act (CCPA) stands out as a representative regulatory framework, requiring companies to disclose data usage practices more transparently and granting consumers the right to request data deletion [32].

In China, digital platform governance has increasingly shifted from traditional cybersecurity regulation toward broader concerns regarding interface fairness and user autonomy. The 2021 Regulations on the Administration of Algorithmic Recommendations for Internet Information Services explicitly prohibit platforms from using algorithms to induce addiction, manipulate behavior, or violate public order, and ensure that users must retain the right to disable personalized recommendations [33]. Although this regulation does not directly use the term “dark patterns,” it introduces preliminary constraints on interface-based behavioral manipulation. Meanwhile, the Personal Information Protection Law, Consumer Rights Protection Law, and Data Security Law all establish clear protections for user informed consent, data autonomy, and choice rights [34].

However, existing legal frameworks—both domestic and international—often remain principle-oriented and lack clear definitions, technical standards, and operative criteria specific to dark patterns. This leads to ambiguity and enforcement delays in real-world regulation [35]. For example, although GDPR imposes strict consent requirements, it does not clearly define what constitutes a dark pattern, limiting enforcement precision.

To address these challenges, several legal governance strategies are necessary:

Clarification of legal terminology and definitions. Regulatory frameworks should explicitly define “dark pattern design” and its specific variants to enhance enforceability [36][37]. This process may be supported by the adoption of interface transparency standards such as ISO 9241-210 [38].

Establishment of pre-implementation assessment mechanisms. Regulation should shift from post-hoc punishment to pre-event review. Since dark pattern manipulation often occurs at a subconscious level during user interaction, relying on ex-post accountability cannot effectively restore user losses. Platforms—especially in high-risk domains such as finance, healthcare, and data-intensive services—should be required to undergo interface ethics risk assessments prior to deployment [35].

Strengthening user empowerment mechanisms. This includes requiring platforms to adopt clear language, simplified visual structures, and parallel cancellation paths, while also developing digital literacy education to enhance users’ ability to recognize manipulation, evaluate interface cues, and make autonomous decisions.

Within the digital platform governance system, ethical measures are increasingly becoming an important complement to legal regulation. Several countries have begun exploring institutionalized ethical constraints on dark patterns. For example, the United Kingdom’s Information Commissioner’s Office introduced the Age Appropriate Design Code, which, based on the protection of minors’ rights, requires interface designs to avoid default push mechanisms and prohibits the use of encouraging or manipulative language, thereby emphasizing respect for user autonomy [39]. Similarly, in 2022, the New Zealand Privacy Commissioner launched the Digital Ethics Lab initiative, promoting interdisciplinary collaboration to evaluate platform design ethics and requiring platforms to explain the intentions behind interface design and its influence on user behavior [40]. These policy developments show that ethical governance should not focus solely on post-hoc penalties but should instead emphasize embedded responsibility in the design stage and the preemptive protection of user rights.

In China, ethical governance of interface design is also gaining attention. The Provisions on the Governance of the Online Information Content Ecosystem (2020) require internet service providers to establish mechanisms for human oversight and user autonomy to prevent over-consumption, addiction, and user rights infringement [41]. The Administrative Provisions on Internet User Public Account Information Services (2021) prohibit misleading distribution mechanisms and require platforms to avoid practices such as “click-bait inducement” or false recommendation displays [42]. In addition, the 2023 directive Notice on Improving Mobile Internet Application Service Capabilities mandates that services such as automatic renewal must not be pre-selected or force-bundled, and that applications must provide clear and accessible exit paths [43]. Although these policies do not explicitly use the term dark patterns, they already address core issues such as inducement-based design, technical manipulation, and forced bundling.

However, both domestic and international ethical norms in interface design have historically focused more on commercial platforms and have not sufficiently addressed public services and social applications. Existing frameworks tend to remain at the macro-principle level, with limited focus on concrete design practices, algorithmic mechanisms, and differentiated effects across groups. Furthermore, the ethical implications of dark patterns vary significantly across social groups, disproportionately affecting the elderly, adolescents, individuals with lower educational attainment, and linguistic minorities.

Given these challenges, ethical governance of dark patterns should include the following measures:

First, introduce ethical risk assessment systems across multiple digital platform domains. Before deployment, digital products—beyond commercial platforms and including public and social service systems—should undergo a behavioral impact assessment, conducted by multidisciplinary ethics review committees that evaluate manipulative potential, risk exposure, and user controllability.

Second, strengthen explainable interface design. Platforms should provide clear semantics, design rationales, and risk explanations regarding interface and algorithmic operations. Moreover, before the platform is put into use, independent evaluators should assess whether interface information is cognitively accessible to users of different ages and educational backgrounds [44].

Third, promote standardization of ethical norms. International standards bodies (e.g., ISO, W3C) or national regulatory agencies could develop a Digital Interface Design Ethics Standard to regulate practices including default selections, hidden rejection pathways, and emotionally manipulative language [45].Ethical design education and professional codes of conduct—such as those promoted by ACM SIGCHI—should be integrated into designer training to strengthen accountability at the source of interface creation [46].

In international contexts, the technical governance of dark pattern design has increasingly emphasized standardization and tool-based detection, with the primary aim of improving recognition capabilities and supporting corresponding legal and ethical oversight. For example, researchers have proposed the Dark Pattern Analysis Framework, which categorizes and standardizes existing design strategies and identifies 68 distinct types. However, current detection tools are capable of identifying fewer than half of these types, and the scope of existing datasets remains limited—suggesting that the complexity of dark patterns still exceeds current technical recognition capabilities [47]. Similarly, some scholars have introduced artificial intelligence into interface analysis. The AidUI system integrates computer vision and natural language processing to detect more than ten forms of dark patterns in interface screenshots [48]. The AppRay system draws upon a large repository of dark pattern cases and demonstrates strong potential in detection performance [49].

In China, increasing attention is also being paid to the identification and detection of dark pattern design. Scholars have systematically classified dark patterns in e-commerce and financial platforms, providing conceptual foundations for future detection tools. Furthermore, during the 2021 nationwide technical inspection, the Ministry of Industry and Information Technology reported that several mobile applications contained “forced permissions” and “path obstacles,” and subsequently issued public notices or removal orders against non-compliant platforms [50]. Research and development of automated detection systems for dark patterns are currently underway.

However, with the advancement of affective computing, immersive interface design, and generative AI, interface manipulation is becoming increasingly micro-scale and adaptive, raising the difficulty of technological governance. To address these challenges, several strategies should be adopted:

First, develop interaction assistance tools and dark pattern detection systems. Platforms may design browser extensions or interface overlays that highlight default settings, path barriers, or visual inducement elements in real time. Additionally, generative AI and natural language processing models may be employed to construct automated detection frameworks, transforming algorithmic capabilities into cognitive support tools for users. For example, Mills and Whittle (2023) propose a detection system based on language generation models capable of identifying visual emphasis, terminological ambiguity, and procedural obstruction in interface design [51].

Second, promote multi-stakeholder co-governance by integrating technical evaluation, public oversight, and ethical deliberation. This may include establishing multidisciplinary design review committees involving user representatives, ethicists, and legal advisors to conduct behavioral impact assessments [52].

Third, strengthen public digital literacy education. Educational and public service systems should train users to recognize dark pattern strategies and interface manipulation techniques, improving their sensitivity to and resistance against technological influence [53][54]. Practical challenges remain, including the rapid evolution of design technologies, cross-jurisdictional inconsistencies, and contextual variations in manipulation perception.

Dark pattern design is a specialized form of interface manipulation driven by algorithms and behavioral guidance mechanisms. By leveraging default settings, visual cues, and path architecture, digital platforms construct a decision-making environment that appears voluntary yet is covertly controlled. This design has widespread implications across e-commerce, finance, education, healthcare, and social interaction domains, resulting in infringements on user interests, cognitive interference, and privacy exposure.

Therefore, a comprehensive governance framework must be established combining legal, ethical, and technical measures. Key governance directions include: Clarifying legal definitions; Establishing ex-ante evaluation mechanisms; Enhancing user empowerment; Implementing ethical risk review systems; Strengthening explainable interface design; Promoting ethical standardization; Developing automated detection tools; Supporting participatory design governance; And advancing public digital literacy education. These efforts together will support the construction of a fair, transparent, and user-centered digital society.