The Influence of AI Chatbot Usage Frequency and Digital Behavior on Information Ethics Compliance among Indonesia’s Digital Society

The rapid development of artificial intelligence (AI), particularly generative chatbots such as ChatGPT, has significantly transformed the way individuals access and share information in the digital era. In Indonesia, the use of this technology has become widespread across various segments of society, including students, learners, and librarians. This has created a dynamic yet complex information ecosystem, with challenges related to adherence to information ethics, such as plagiarism and academic integrity. Research shows that while generative AI offers convenience in accessing information, unethical use can contribute to violations of academic norms as well as intellectual integrity and authenticity(Ouchchy et al., 2020).In this context, concerns have emerged that the adoption of generative AI may influence users’ ethical behavior, creating an urgent need for awareness of information literacy. Previous studies indicate that limited understanding of digital ethics and information literacy among students and learners may lead to unethical behavior in the use of technology(Harmanto et al., 2022; Prasetiyo et al., 2021). Other studies have also highlighted that the use of generative AI can affect users’ ethical conduct. For instance,Ali & Aysan (2025)found that in academic settings, maintaining intellectual authenticity and integrity poses significant challenges due to the use of generative AI. However, research that specifically examines the relationship between the frequency of chatbot use, digital behavior, and adherence to information ethics in Indonesia remains limited. This gap underscores the need for further in-depth studies to better understand these dynamics within the context of Indonesia’s digital society. As reliance on AI continues to increase, it is essential for society to be equipped with adequate knowledge and skills to navigate the abundance of information circulating in the digital world.To analyze this phenomenon, several theoretical frameworks have been applied. For example, the Theory of Planned Behavior (TPB) byAjzen (1991)emphasizes how attitudes, subjective norms, and perceived behavioral control can influence individuals’ intentions to use AI ethically(Kumari et al., 2023). In addition, the Information Ethics approach byCapurro (2006)highlights the moral responsibility of individuals in accessing information(Ouchchy et al., 2020). Ribble (2011)framework of Digital Citizenship is also useful in evaluating digital behavior and ethics in online contexts, stressing the importance of social responsibility(Prasetiyo et al., 2021). Collectively, these frameworks underscore the importance of digital literacy and education in information ethics to address the complex challenges of Indonesia’s digital society.A clear research gap emerges as only a few studies have comprehensively examined the relationship between AI chatbot usage frequency, digital behavior, and adherence to information ethics, particularly within the context of Indonesia’s digital society. Existing research has largely focused on descriptive approaches or case studies with limited scope and has not sufficiently employed strong theory-based quantitative methods. This opens up opportunities to design a study that bridges these three variables within a cohesive conceptual framework.To address this gap, the Theory of Planned Behavior(Ajzen, 1991)will be applied to explain how ethical intentions and behaviors are shaped by attitudes, subjective norms, and perceived behavioral control in the use of AI chatbots. The Information Ethics framework (Capurro, 2006)will assist in analyzing individuals’ moral responsibility in utilizing digital information ethically. Meanwhile, the Digital Citizenship framework(Ribble, 2011)will be employed to measure the extent to which individuals’ digital behavior aligns with the values of ethics, responsibility, and technological literacy.Building on this understanding, the proposed study aims to address questions concerning the impact of AI chatbot usage frequency and digital behavior on adherence to information ethics among Indonesia’s digital society. This research is expected to providedeeper insights into the relationship between AI use, digital behavior, and ethical compliance in information use, as well as recommendations for strengthening information ethics literacy in today’s digitalized era</p>

This study employs a descriptive quantitative approach aimed at identifying the influence of AI chatbot usage frequency and digital behavior on information ethics compliance among Indonesia’s digital society. The research was conducted over one month, fromthe beginning to the end of May 2025. Data were collected through an online survey distributed via Google Forms across various social media channels and digital communities, such as WhatsApp Groups, Telegram Channels, and interest-based platforms (e.g., AI communities, student groups, and digital learning forums).The sampling technique applied was non-probability sampling with an accidental sampling approach, in which any individual who accessed and voluntarily completed the questionnaire during the data collection period was included as a respondent. A total of 458 valid responses were obtained. The research instrument consisted of three main sections: (1) AI chatbot usage frequency (5 items), (2) digital behavior indicators (5 items), and (3) level of compliance with information ethics (5 items). Each item was measured using a 5-point Likert scale, ranging from “strongly disagree” to “strongly agree.”Data analysis was conducted using PLS-SEM with SmartPLS version 3.9. The analysis stages began with the outer model assessment to test indicator validity and construct reliability, including convergent validity (factor loadings and AVE), discriminant validity, and composite reliability. This was followed by the inner model assessment to examine the relationships among latent variables using path coefficients, R-square values, effect size (f²), and predictive relevance (Q²). The evaluation model also included bootstrapping to test the significance of the structural paths at a 0.05 significance level.SmartPLS was selected as the analytical tool because the proposed model is predictive and exploratory in nature and is suitable for testing theoretical models with multiple latent constructs and data that do not need to follow a normal distribution.</p> <p>In addition, PLS-SEM is effective for research in dynamic digital social contexts such as technology behavior and information ethics, which often involve heterogeneous samples.The Theory of Planned Behavior(Ajzen, 1991)was adopted as the conceptual framework because it is relevant for explaining how individual behavior (in this case, technology use) is influenced by attitudes, subjective norms, and perceived behavioral control. This theory supports the exploration of the relationship between the intensity of chatbot usage and adherence to information ethics norms in the digital sphere. Furthermore, the digital ethics elements from Ribble (2011)digital citizenship model were incorporated to measure information ethics compliance as part of digital responsibility. The integration of these two theories enables a comprehensive understanding of information ethics compliance as an outcome of cognitiveand social processes within the digital environment.Figure 1. Conceptual Framework of the StudySource: (Ajzen, 1991; Capurro, 2006; Ribble, 2011)The research framework was designed to analyze the influence of AI chatbot usage frequency (X1) and digital behavior (X2) on information ethics compliance (Y) among Indonesia’s digital society. It is based on the assumption that the more frequently individuals interact with AI chatbots, the greater their exposure to information requiring ethical responsibility in its use. At the same time, digital behavior—including how individuals interact online, their awareness of privacy, and their level of information literacy—is also predicted to play an important role in encouraging compliance with information ethics.The relationships among these three variables are illustrated in a conceptual model that depicts the direct effects of X1 and X2 on Y. This model is tested using a quantitative approach through path analysis with PLS-SEM, implemented in SmartPLS. This approach allows the researcher to assess causal relationships among variables and evaluate the predictive power of each path in the model.Based on this conceptual framework, the study formulates three main hypotheses: H1: There is a positive and significant influence of AI chatbot usage frequency on information ethics compliance; H2: There is a positive and significant influence of digital behavior on information ethics compliance; andH3: There is a AI chatbot usage frequency(X1)Digital Behavior(X2)Information Ethics Compliance(Y)H1H2H3 simultaneous positive and significant influence of AI chatbot usage frequency and digital behavior on information ethics complianceThese hypotheses will be tested to determine the extent to which each independent variable contributes to enhancing individuals’ compliance with principles of information ethics, which has become a critical issue in the increasingly AI-driven digital era. Through this framework, the study is expected to provide both theoretical and practical contributions to the development of digital literacy and information ethics policies in Indonesia</p>

This study employs a descriptive quantitative approach aimed at identifying the influence of AI chatbot usage frequency and digital behavior on information ethics compliance among Indonesia’s digital society. The research was conducted over one month, fromthe beginning to the end of May 2025. Data were collected through an online survey distributed via Google Forms across various social media channels and digital communities, such as WhatsApp Groups, Telegram Channels, and interest-based platforms (e.g., AI communities, student groups, and digital learning forums).The sampling technique applied was non-probability sampling with an accidental sampling approach, in which any individual who accessed and voluntarily completed the questionnaire during the data collection period was included as a respondent. A total of 458 valid responses were obtained. The research instrument consisted of three main sections: (1) AI chatbot usage frequency (5 items), (2) digital behavior indicators (5 items), and (3) level of compliance with information ethics (5 items). Each item was measured using a 5-point Likert scale, ranging from “strongly disagree” to “strongly agree.”Data analysis was conducted using PLS-SEM with SmartPLS version 3.9. The analysis stages began with the outer model assessment to test indicator validity and construct reliability, including convergent validity (factor loadings and AVE), discriminant validity, and composite reliability. This was followed by the inner model assessment to examine the relationships among latent variables using path coefficients, R-square values, effect size (f²), and predictive relevance (Q²). The evaluation model also included bootstrapping to test the significance of the structural paths at a 0.05 significance level.SmartPLS was selected as the analytical tool because the proposed model is predictive and exploratory in nature and is suitable for testing theoretical models with multiple latent constructs and data that do not need to follow a normal distribution.</p> <p>In addition, PLS-SEM is effective for research in dynamic digital social contexts such as technology behavior and information ethics, which often involve heterogeneous samples.The Theory of Planned Behavior(Ajzen, 1991)was adopted as the conceptual framework because it is relevant for explaining how individual behavior (in this case, technology use) is influenced by attitudes, subjective norms, and perceived behavioral control. This theory supports the exploration of the relationship between the intensity of chatbot usage and adherence to information ethics norms in the digital sphere. Furthermore, the digital ethics elements from Ribble (2011)digital citizenship model were incorporated to measure information ethics compliance as part of digital responsibility. The integration of these two theories enables a comprehensive understanding of information ethics compliance as an outcome of cognitiveand social processes within the digital environment.Figure 1. Conceptual Framework of the StudySource: (Ajzen, 1991; Capurro, 2006; Ribble, 2011)The research framework was designed to analyze the influence of AI chatbot usage frequency (X1) and digital behavior (X2) on information ethics compliance (Y) among Indonesia’s digital society. It is based on the assumption that the more frequently individuals interact with AI chatbots, the greater their exposure to information requiring ethical responsibility in its use. At the same time, digital behavior—including how individuals interact online, their awareness of privacy, and their level of information literacy—is also predicted to play an important role in encouraging compliance with information ethics.The relationships among these three variables are illustrated in a conceptual model that depicts the direct effects of X1 and X2 on Y. This model is tested using a quantitative approach through path analysis with PLS-SEM, implemented in SmartPLS. This approach allows the researcher to assess causal relationships among variables and evaluate the predictive power of each path in the model.Based on this conceptual framework, the study formulates three main hypotheses: H1: There is a positive and significant influence of AI chatbot usage frequency on information ethics compliance; H2: There is a positive and significant influence of digital behavior on information ethics compliance; andH3: There is a AI chatbot usage frequency(X1)Digital Behavior(X2)Information Ethics Compliance(Y)H1H2H3 simultaneous positive and significant influence of AI chatbot usage frequency and digital behavior on information ethics complianceThese hypotheses will be tested to determine the extent to which each independent variable contributes to enhancing individuals’ compliance with principles of information ethics, which has become a critical issue in the increasingly AI-driven digital era. Through this framework, the study is expected to provide both theoretical and practical contributions to the development of digital literacy and information ethics policies in Indonesia</p> construct, X1*X2, represents the interaction between AI Chatbot Usage Frequency and Digital Behavior, with high loading factor values (0.790 to 0.895).The dependent variable, Information Ethics Compliance (Y), is measured through five indicators (Y1 to Y5), with loading factor values ranging from 0.844 to 0.877, reflecting high measurement reliability. From the structural paths, it can be observed that: AI Chatbot Usage Frequency (X1) has a direct negative effect on Information Ethics Compliance (Y) with a coefficient of -0.257. Digital Behavior (X2) exerts a positive influence on Information Ethics Compliance (Y) with a coefficient of 0.219. The interaction term (X1*X2) has a very strong and positive effect on Information Ethics Compliance (Y), with a coefficient of 0.953.In PLS model testing, two main aspects of validity must be considered: convergent validity and discriminant validity(Purnomo, 2019). Convergent validity assesses the extent to which indicators accurately represent the intended construct, while discriminant validity tests whether a construct is truly distinct and not overlapping with other constructs. These validity tests are crucial in the process of model development and validation in quantitative research. Based on the data processed using the PLS algorithm as shown in Figure 2, all indicators in the model meet the required validity standards. Referring toWiyono (2011), loading factor values between 0.50 and 0.60 are considered adequate to demonstrate convergent validity, and all indicators in this study exceeded that threshold. Furthermore, discriminant validity was tested using the cross-loading method. As noted byHair et al. (2014), an indicator is considered discriminantly valid if its highest cross-loading value appears on the construct it is intended to measure compared to other constructs. When this pattern is consistent, the model is deemed to meet discriminant validity and can proceed to subsequent testing stages, as detailed in Table 1.Table1. Results of the AlgoritmTestVariableAVEComposite ReliabilityR Square (R2)RemarksAI Chatbot Usage Frequency (X1)0.5790.872ReliableDigital Behavior (X2)0.6460.901ReliableInformation Ethics Compliance (Y)0.7400.9340.863ReliableX1*X20.7180.9270.972ReliableSource: SmartPLS 3.9 OutputBased on the algorithm testing results presented in Table 1, all variables in the research model demonstrate high reliability and have met the criteria for convergent validity. This is evidenced by the Average Variance Extracted (AVE) values, all of which exceed 0.5, meaning that more than 50% of the variance in the indicators can be explained by their respective constructs. The AI Chatbot Usage Frequency (X1) variable achieved an AVE of 0.579 and a Composite Reliability (CR) of 0.872, indicating that the construct is reliable and consistent in measuring the intended variable. Similarly, the Digital Behavior (X2) variable recorded an AVE of 0.646 and a CR of 0.901, demonstrating strong internal consistency.The dependent variable, Information Ethics Compliance (Y), recorded an AVE of 0.740, a CR of 0.934, and an R Square (R²) value of 0.863. This R² value indicates that the combination of X1 and X2 explains 86.3% of the variance in Y, demonstrating very strong predictive power. Furthermore, the interaction construct (X1*X2) also displayed strong reliability with an AVE of 0.718, a CR of 0.927, and an R² of 0.972. This confirms that the interaction variable makes a significant contribution to explaining changesin Y. Overall, these findings show that all constructs in the model meet the requirements of validity and reliability, allowing the research to proceed to the next stage of analysis.To assess the predictive relevance of a model, the Q-square (Q²) statistic is used. This measure evaluates how well the model and its estimated parameters are able to predict observed values. According toGhozali (2008), if the Q-square (Q²) value is greater than zero, the model is considered to have relevant predictive ability (predictive relevance). Conversely, a Q² value below zero indicates insufficient predictive power. The formula for calculating Q-square (Q²) is as follows:𝑄2=1−(√1−𝑅12)×(√1−𝑅22)Applying this calculation:𝑄2=1−(√1−𝑅12)×(√1−𝑅22)𝑄2=1−(√1−0.8632)×(√1−0.9722)𝑄2=1−(√1−0.744769)×(√1−0.944784)𝑄2=1−(√0.255231)×(√0.055216)𝑄2=0.8814The Q² calculation result of 0.8814 indicates that the model has very strong predictive relevance. According to general criteria in SmartPLS analysis, a Q² value above 0.35 is categorized as good predictive relevance. In general, a Q² value greater than zero signifies that the model has strong predictive ability. In the context of this study, the Q² value of 0.8814 demonstrates that the structural model possesses strong predictive relevance, particularly in explaining the dependent variable, Information Ethics Compliance. This finding aligns withGhozali (2008), assertion that a Q-square value greater than zero indicates that the model is capable of adequately predicting observed values. Therefore, the model in this study can be considered to have high predictive quality, especially in explaining the dependent variable. These results thus support the structural model’s validity and show that the independent variables used in the study effectively predict the dependent variable under investigation.</p> <p>Table 2. Results of the Bootstrapping Test for Direct EffectsOriginal Sample(O)Sample Mean(M)Standard Deviation(STDEV)T-Statistics(| O/STDEV |)P-ValuesFrequency of Chatbot AI Use (X1) →Information Ethics Compliance (Y)-0.257-0.2250.1751.4670.143Frequency of Chatbot AI Use (X1) →X1*X20.5860.5860.02127.9720.000Digital Behavior (X2) →Information Ethics Compliance (Y)0.2190.2340.1111.9810.048Digital Behavior (X2) →X1*X20.4760.4780.02618.0670.000X1*X2 →Information Ethics Compliance (Y)0.9530.9110.2413.9490.000Source: SmartPLS 3.9 OutputBased on the results of the bootstrapping test for direct effects presented in Table 2, several key findings emerge regarding the relationships between variables in the research model. This test aims to examine the significance of each path coefficient by analyzing the t-statisticsand p-values.First, the relationship between the frequency of chatbot AI use (X1) and information ethics compliance (Y) shows an original sample value of -0.257, with a t-statisticof 1.467 and a p-valueof 0.143. Since the p-valueis greater than 0.05, this relationship is statistically insignificant. This indicates that, directly, the frequency of chatbot AI use does not exert a strong influence on information ethics compliance. In AI Chatbot Usage Frequency (X1)Digital Behavior (X2)Information Ethics Compliance (Y) other words, the intensity of AI usage alone does not guarantee greater ethical awareness in managing information.Conversely, the relationship between digital behavior (X2) and information ethics compliance (Y) shows a coefficient of 0.219, with a t-statisticof 1.981 and a p-valueof 0.048. Since p < 0.05, this influence is statistically significant. This finding highlights that the more positive one’s digital behavior—such as respecting copyright, maintaining ethical communication, and avoiding disinformation—the higher their compliance with principles ofinformation ethics.Furthermore, both X1 →X1X2 and X2 →X1X2 demonstrate highly significant relationships, with t-statisticsof 27.972 and 18.067, respectively, and p-valuesof 0.000. This confirms that the interaction construct between frequency of AI use and digital behavior is validly established within the model.Most notably, the effect of the interaction term X1*X2 on information ethics compliance is strongly significant, with a coefficient of 0.953, a t-statisticof 3.949, and a p-valueof 0.000. This finding indicates that when the frequency of chatbot AI use is accompanied by positive digital behavior, the effect on information ethics compliance becomes exceptionally strong. The interaction effect is substantially greater than the direct effects of the independent variables.The results of the bootstrapping test for indirect effects (presented in Table 3) further reveal that all indirect relationships in the model are statistically significant. This test examines how the frequency of chatbot AI use (X1) and digital behavior (X2) influence information ethics compliance (Y) when mediated by the interaction construct X1*X2.The indirect effect from frequency of chatbot AI use (X1) on information ethics compliance (Y) shows a coefficient of 0.559, with a t-statisticof 3.979 and a p-valueof 0.000. These results suggest that although the direct effect of X1 on Y is not significant (as shown in Table 2), its indirect effect through the interaction mechanism X1*X2 is both significant and substantial. Thus, the frequency of chatbot AI use will positively impact information ethics compliance only when supported by another critical factor—positive digital behavior.Table3. Results of the Bootstrapping Test for Indirect EffectsOriginal Sample(O)Sample Mean(M)Standard Deviation(STDEV)T-Statistics(|O/STDEV|)P-ValuesFrequency of Chatbot AI Use (X1) →Information Ethics Compliance (Y)0.5590.5330.1403.9790.000Frequency of Chatbot AI Use (X1) →X1*X20.5590.5330.1403.9790.000Digital Behavior (X2) →Information Ethics Compliance (Y)0.4540.4360.1203.7670.000Digital Behavior (X2) →X1*X20.4540.4360.1203.7670.000X1*X2 →Information Ethics Compliance (Y)0.9530.9110.2413.9490.000Source: SmartPLS 3.9 Output Similar findings are observed in the indirect path from digital behavior (X2) to information ethics compliance (Y), with a coefficient of 0.454, a t-statisticof 3.767, and a p-valueof 0.000. This indicates that digital behavior not only has a direct impact but also provides an additional strong effect when combined with the intensity of chatbot AI use. In other words, digital behavior acts as a reinforcer or facilitator that strengthens the relationship between AI use and adherence to ethical principles.Meanwhile, the path coefficient for X1*X2 →Y remains very high at 0.953, with a t-statisticof 3.949 and a p-valueof 0.000. This shows that the interaction construct between AI use and digital behavior exerts the strongest influence on information ethics compliance compared to other paths. This finding underscores the importance of synergy between technology use and digital behavioral traits in shaping ethical attitudes toward information.Overall, the results of the indirect effect test reinforce the view that ethical behavior in the use of AI cannot be shaped solely by the intensity of technology use but depends heavily on how individuals behave digitally as a whole. Therefore, educationalinterventions or policy measures aimed at improving information ethics compliance should focus on developing responsible digital behavior, not merely on enhancing technical proficiency in AI usage.The results of the bootstrapping test for total effects, as presented in Table 4, provide a comprehensive overview of the strength of relationships between variables, covering both direct and indirect influences on information ethics compliance. All paths analyzed are statistically significant, with p-valuesbelow 0.05, and most at the 0.000 level, indicating very strong significance. The total path from the frequency of chatbot AI use (X1) to information ethics compliance (Y) shows an original sample valueof 0.302, with a t-statisticof 4.433 and a p-valueof 0.000. This result indicates that although the direct effect was not significant in earlier tests, when combined with indirect effects, the total effect of X1 on Y becomes significant. This strengthens the argument that AI use can contribute to information ethics compliance, particularly when supported by another key factor: ethical digital behavior.Table4. Results of the Bootstrapping Test for Total Effects (Direct and Indirect)Original Sample(O)Sample Mean(M)Standard Deviation(STDEV)T-Statistics(|O/STDEV|)P-ValuesFrequency of Chatbot AI Use (X1) →Information Ethics Compliance (Y)0.3020.3080.0684.4330.000Frequency of Chatbot AI Use (X1) →X1*X20.5860.5860.02127.9720.000Digital Behavior (X2) →Information Ethics Compliance (Y)0.6730.6690.06510.3790.000Digital Behavior (X2) →X1*X20.4760.4780.02618.0670.000X1*X2 →Information Ethics Compliance (Y)0.9530.9110.2413.9490.000Source: SmartPLS 3.9 Output Meanwhile, the total effect of digital behavior (X2) on information ethics compliance is higher, with a coefficient of 0.673, a t-statisticof 10.379, and a p-valueof 0.000. This confirms that digital behavior plays a major role in influencing overall information ethics compliance. This result is consistent with the direct and indirect effect tests, in which X2 consistently showed a significant impact on Y. The interaction effect of X1*X2 on Y also remains highly significant, with a coefficient of 0.953 and a t-statisticof 3.949. This indicates that the combined effect of chatbot AI usage frequency and digital behavior produces the strongest impact on information ethics compliance. The interaction illustrates that intensive technology use will only yield positive outcomes if carried out by individuals who demonstrate responsible digital behavior.In conclusion, these findings reinforce the theoretical model applied in this study: the combination of positive digital behavior and optimal AI use results in a significant increase in compliance with information ethics. This model is particularly relevant in the context of Indonesia’s digital society, which is rapidly expanding its use of AI but still faces major challenges in digital ethics literacy. Therefore, strengthening digital behavior emerges as the key to ensuring that technological adoption contributes ethically and responsibly.Hypothesis 1: There is a positive and significant effect of chatbot AI usage frequency on information ethics complianceBased on the results of the bootstrapping test on the total effect presented in Table 4, it was found that the frequency of chatbot AI usage (X1) has a positive and significant effect on information ethics compliance (Y), with a coefficient value of 0.302,a t-statistic of 4.433, and a p-value of 0.000. This finding indicates that the more frequently individuals use chatbot AI, the higher their level of compliance with information ethics. In this context, users’ knowledge and awareness of ethical issues such as data privacy and information transparency become highly important. This aligns with the findings ofSuharmawan (2023), who stated that the use of technology, including ChatGPT, in educational contexts requires careful attention to privacy and ethical aspects.However, it is important to note that the positive influence of chatbot AI usage frequency is not automatic. Ethical usage requires a deep understanding and awareness of the principles of information ethics. Research byPrasetyaningrum et al. (2022)explained that morality and social issues influence ethics in information systems, showing that the social context and individual morality play a role in ethical decision-making related to technology use.The importance of providing education on AI ethics was emphasized byHamsar et al. (2024), who found that students have a solid understanding of ethical aspects in AI use, demonstrating that ethical competence can contribute to responsible decision-making in technology use (Piispanen et al., 2024; Stöhr et al., 2024). Thus, this study reinforces that in order to improve information ethics compliance through chatbot AI usage, it is strongly recommended to educate users about ethical principles and social responsibility in technology.Hypothesis 2: There is a positive and significant effect of digital behavior on information ethics complianceBased on the bootstrapping test results on the total effect presented in Table 4, it was found that digital behavior (X2) has a positive and significant effect on information ethics compliance (Y), with a coefficient value of 0.673, a t-statistic of 10.379, and a p-value of 0.000. This finding supports Hypothesis 2, which states that digital behavior has a positive and significant effect on information ethics compliance.This positive influence indicates that individuals with ethical digital behavior tend to have higher levels of compliance with principles of information ethics. This can be explained by the increasing awareness among users regarding ethical aspects of digital technology use, such as data privacy, information transparency, and responsibility in information dissemination. Individuals with ethical digital behavior are more sensitive to issues of data privacy, information transparency, and responsible conduct in digital spaces(Piispanen et al., 2024).Good digital behavior is often reflected in ethical social media use and online interactions. Research by Rahman et al. (2023)found that social media use positively correlates with how individuals ethically seek academic information. Similarly,Saputri et al. (2024)highlighted the importance of digital literacy in shaping students’ information-seeking behavior, contributing to more ethical decision-making. These results underscore that strong digital literacy plays a crucial role in fostering compliance with information ethics through ethical digital behavior in a rapidly changing information era.Nevertheless, it is important to emphasize that ethical digital behavior does not emerge without adequate education and understanding.Hidayah (2018)highlighted that a good understanding of digital ethics and technology use can reduce unethical digital behavior.Arochma et al. (2023)also demonstrated the importance of ethics education in the context of information technology use, especially with emerging technologies such as ChatGPT. Overall, this study indicates that improving ethical digital behavior and strengthening understanding of ethical information principles play a major role in achieving higher compliance with information ethics, which is particularly crucial amid the rapid development of digital technologies today.Hypothesis 3: There is a simultaneous positive and significant effect of chatbot AI usage frequency and digital behavior on information ethics complianceBased on the bootstrapping test results on the total effect presented in Table 4, it was found that chatbot AI usage frequency (X1) and digital behavior (X2) simultaneously have a positive and significant effect on information ethics compliance (Y). The interaction coefficient between X1 and X2 on Y is 0.953, with a t-statistic of 3.949 and a p-value of 0.000, showing that the combination of intensive chatbot AI usage and good digital behavior jointly enhances individuals’ compliance with principles of information ethics. This finding indicates that intensive chatbot AI use combined with responsible digital behavior can significantly strengthen users’ compliance with ethical standards in information.A study byPrastyono et al. (2023), which emphasized the influence of chatbot AI usage on user character, revealed that appropriate education on technology use can foster better ethical understanding. With users who combine chatbot AI utilization and ethical digital behavior, individuals aremore likely to value privacy, transparency, and responsibility in information sharing, thereby creating a more ethical digital environment.Moreover,Anggadini (2019)stressed that digital literacy and awareness of information ethics are crucial when using new technologies, including chatbot AI. The findings of this study support the idea that increased adoption of such technologies must be balanced with the developmentof ethical understanding.Shellen et al. (2023)further highlighted the importance of developing chatbots that consider ethics and security, emphasizing that technical aspects are critical to building ethical interactions with users.However, it should be noted that this positive influence does not occur automatically. This underlines the importance of education and training that not only focus on technical aspects of AI utilization but also on the development of ethical digital behavior.Hidayah (2018), emphasized the need for a clear understanding of technology ethics, showing that comprehensive education supports individuals in acting ethically when using digital technologies. Overall, the results of this study support the view that the combination of chatbot AI usage frequency and ethical digital behavior significantly contributes to information ethics compliance, which is increasingly vital in today’s digital era.</p>

This study demonstrates that both AI chatbot usage frequency and digital behavior significantly influence information ethics compliance among Indonesia’s digital society. While frequent chatbot use contributes positively, its effect becomes stronger when mediated by ethical digital behavior. The findings highlight that responsible digital behavior acts as the key driver in linking AI utilization with ethical information practices. Theoretically, this research extends the application of PLS-SEM in explainingthe relationship between AI adoption and ethical compliance, emphasizing the mediating role of digital behavior. Practically, the results underscore the urgent need for integrating digital literacy and ethics education into both formal and non-formal learning contexts. Governments, educational institutions, and technology developers should collaborate to design policies and training programs that combine technical skills with ethical values, ensuring that AI-driven information ecosystems foster responsible and sustainable digital citizenship</p>