Fostering social media user intentions: AI-enabled privacy and intrusiveness concerns

Purpose – This paper aims to empirically examine the impact of psychological factors (i.e. privacy and intrusiveness concerns) on user intentions regarding arti ﬁ cial intelligence (AI)-enabled social commerce applications at their core through perceived usefulness. The theoretical model is supported by the theory of planned behaviour (TPB). Design/methodology/approach – Data was gathered from488socialmedia usersin Saudi Arabia. Findings – Privacy concerns signi ﬁ cantly affect perceived usefulness. Furthermore, the link between privacy concerns andbehavioural intentionswasmediated by perceived usefulness. Research limitations/implications – Business leaders should raise users ’ awareness about the effectiveness ofAI-powered toolsthat can in ﬂ uencetheir behaviouralintentions. Furthermore, managersmust be awareoftheregulationsthatprotectuserprivacy,trackonlineactivityandoffersecurecommunicationchannels.


Introduction
Artificial intelligence (AI) is currently essential as businesses move towards a more algorithmic environment (Soliman et al., 2023).Technology is becoming a participant in the value exchange process rather than just a tool used by stakeholders (Aguirre et al., 2023).While decreasing human interaction with technology, this shift in AI draws the attention of practitioners (Anwar et al., 2023).The majority of industries, including retail, advertising and e-commerce, adopt AI-powered solutions (Blanco-Moreno et al., 2023).These apps enable users to communicate through digital interfaces by using natural-language computer programmes that approximate human speech (Selem et al., 2023c).
Despite their importance in a variety of industries, AI-driven apps face challenges.Because AI suggests that technology owners will have greater access to and control over private user data (i.e.demographic, geographic and activity characteristics), not all stakeholders may support them (Soliman et al., 2023).Interactions between humans and AI generate a large amount of behavioural and personal data.Data is critical to the success and development of AI, as it influences user behavioural goals.When using digital help, people willingly contribute to the loss of their privacy (Selem et al., 2023a).In a knowledge-based digital environment, algorithms, the internet and data raise privacy and intrusiveness problems (Soliman et al., 2023;Nazneen et al., 2023).
Previous research has examined the potential impact of privacy concerns on promoting behavioural intentions (Da Veiga, 2022;Shoukat et al., 2023a).The psychological variables underlying AI-based applications significantly affect behavioural intentions (Gutierrez et al., 2023;Selem et al., 2023d).These variables are important behavioural intention antecedents in human-technology interaction (Elgammal and Al-Modaf, 2023).Negative reinforcements, such as privacy and intrusiveness concerns, can influence behavioural intentions (Shoukat et al., 2023a).Users' worries about how much access businesses have to personal data are reflected in their concerns about intrusiveness (Selem et al., 2023b(Selem et al., , 2023c)), which impacts their degree of comfort with how businesses use the information they enter on their websites (Selem et al., 2023a).
Otherwise, privacy concerns arise when personal information is made available online (Kronemann et al., 2023;Selem et al., 2023d).This paper investigates perceived usefulness's potential as a mediator between two negative drivers and behavioural intentions.This is one of the few studies that use perceived usefulness as a mediator to examine the pivotal roles of privacy and intrusiveness concerns on purchasing intentions via social media.
This paper contributes to the theory by expanding user privacy research and supporting the TPB framework with AI-powered social commerce platforms used by Saudi Arabian clients.The psychological factors (e.g.privacy and intrusiveness concerns) that influence user adoption of AI-powered social commerce platforms are investigated.The study's aims are met by combining the two fundamental components of TPB: behavioural intentions and perceived usefulness.Rather than focusing on the AI environment, much research has focused on social commerce in general (Abed, 2020;Selem et al., 2023d).
The study contributes to understanding how consumers' buying intentions are influenced by their personal preferences, privacy cues and intrusion fears.By demonstrating the direct effects of perceived benefits on social media user intentions, this research seeks to ascertain the effects of intrusion and privacy concerns on perceived usefulness, which influences behavioural intentions.In Saudi Arabia, the perceived advantages of social media use might be impacted by privacy concerns and interference.Furthermore, how does perceived usefulness function as a mediator between privacy issues and behavioural intention, as well as intrusion concerns and behavioural intention?

Fostering social media user intentions
This study uses theory of planned behaviour (TPB) to support the proposed model.It explores how negative psychological factors influence client purchasing intentions.Despite earlier research in the social commerce scenario focusing on these subjects independently (Elgammal et al., 2023), no study has incorporated these variables into a cohesive framework.Second, we used TPB to explain these notions and experiment with variations in individual behaviour because there is a scarcity of empirical evidence on changes and variations in user behaviour.

Theoretical perspective and hypothesis development
2.1 Perceived usefulness of artificial intelligence-assistance tool According to Soliman et al. (2023), perceived usefulness refers to the degree to which users believe that a given technology would help them perform better in a given work and provide those advantages and benefits.Perceived usefulness, according to Al-Qudah et al. (2022), is the extent to which a user feels certain that a particular good or service will enable them to achieve their objectives.Because of this, perceived usefulness is a crucial component that may influence consumers' decisions to buy (Selem et al., 2023d), and it is crucial to take this into account while thinking about internet technology.Selem et al. (2023a) assert that sentiments towards social media platform usages are greatly affected by websites' perceived usefulness by their users.
The current work intends to evaluate perceived usefulness as a mediation effect in associating negative psychological aspects with behavioural intention in AI since we believe it is a contextual element of TPB (Selem et al., 2023d).The technology acceptance model (TAM) supports this notion, claiming that attitude and individual internal opinions about perceived usefulness work together to establish behavioural intentions (Anwar et al., 2023).Perceived digital usefulness has been established as a predictor of the desire to use mobile apps (Soliman et al., 2023).Thus, this study proposes that: H1.Perceived usefulness of artificial intelligence-assistance tools positively influences behavioural intention.

Intrusiveness and privacy concerns
Intrusiveness concerns are a customer's way of expressing worries about the amount of access that businesses and society have to personal information.As a result, a person's comfort level and the amount of private information that businesses use when posting content on social media are decided (Shoukat et al., 2023a).Such retargeting appears to be causing a feeling of intrusion, based on earlier assessments.Customers' perceptions of this social media intrusion worry and how it affects their behavioural intentions are not well studied (see Lee et al., 2022;Selem et al., 2023a).In contrast, user inventiveness rises when technology is perceived as limiting accessibility.Because behavioural intention to use AIpowered platforms is facilitated by perceived usefulness, the authors think that intrusiveness issues will have a detrimental impact on AI-assistance tools' usefulness.Thus, the following hypotheses are predicted in this paper: H2a. Intrusiveness concerns negatively influence the perceived usefulness of artificial intelligence-assistance tools.
H2b.The perceived usefulness of artificial intelligence-assistance tools mediates the association between intrusiveness concerns and behavioural intention.
Privacy concerns are closely related to online access to personal data (Jaspers and Pearson, 2022).Privacy considerations should not be ignored while making online purchases, according to Selem et al. (2023d), as they have the ability to erode customer trust and behavioural intentions.Conversely, those with strong privacy concerns showed disapproval of direct marketing, which negatively affected their desire to purchase (Higueras-Castillo et al., 2023).Furthermore, privacy issues continue to be a problem that inevitably restricts the development of AI-powered systems (Selem et al., 2023d).
Educating customers about privacy concerns has proven to be the hardest task for merchants.User behaviour literature rarely mentions privacy problems.Alzaidi and Agag (2022), for example, stress how important it will be in the future to look at continuous customer purchases if participants feel comfortable that their privacy will be respected.Furthermore, it's uncertain how consumers would weigh the importance of these privacy concerns in relation to their shopping preferences (Kronemann et al., 2023).
Privacy policies that provide users with control over data collection may be adopted to address privacy concerns (Da Veiga, 2022).Customers who use AI-powered applications indicate growing concern about privacy issues (Alzaidi and Agag, 2022).According to Higueras-Castillo et al. (2023), privacy concerns and perceived usefulness should be examined together.Regardless of how positively or negatively people perceive the benefits of technology; negative psychological difficulties have been identified as potential barriers to AI-enabled enterprises success.Because it forecasts customer behavioural intents, privacy concerns raise uneasiness and have a substantial impact on customers' views towards using AI-powered social commerce platforms (Tseng, 2023).As an increasing phenomenon, privacy concerns should be investigated in AI-related situations (Selem et al., 2023d).Hence, this paper adopts that: H3a.Privacy concerns negatively influence the perceived usefulness of artificial intelligence-assistance tools.
H3b. Perceived usefulness of the artificial intelligence-assistance tool mediates the association between privacy concerns and behavioural intention.

Theoretical framework
The theoretical framework depicted in Figure 1 is based on TPB, a widely established psychological model for understanding and predicting human behaviour, particularly in terms of technology adoption and usage (Jaspers and Pearson, 2022;Morimoto and Macias, 2009).The use of AI support tools in social commerce applications is a study topic in the current case.We used two psychological elements (privacy and intrusiveness concerns) as independent variables in the current theoretical model to determine users' behavioural intentions.Previous researchers have stated that attitude is a significant component determining individual intention to engage in that behaviour (Alzaidi and Agag, 2022;Da Veiga, 2022).These input parameters were selected inside the TPB model's primary tent of "attitude".As a result, privacy concerns have a direct impact on individual attitudes, altering their perceptions of the use of AI help tools.
Similarly, intrusiveness concerns are associated with another important TPB aspect, "perceived behaviour control".TPB believes that concerns about intrusiveness may reduce an individual's impression of control over their interactions with AI technologies, influencing their willingness to use them.Privacy and intrusiveness concerns play an essential role in determining behavioural intentions in the AI context.Furthermore, we used perceived usefulness of AI-powered tools to mediate between privacy concerns and Fostering social media user intentions behavioural intentions, even though this is a well-established TAM mediator (Davis, 1989).By influencing users' perceptions of the benefits of AI products, it serves as a bridge between privacy, intrusiveness concerns and behavioural intention.
To sum up, the literature on technology adoption and psychology that is currently available supports the inclusion of privacy issues, intrusiveness concerns, usefulness and behavioural intention in TPB (Ajzen, 1991).In the context of using AI assistance tools, this theoretical model provides an organised method for comprehending the intricate interactions between various variables.

Measurements
All research items used to evaluate four constructs were adapted from earlier social commerce literature to guarantee content validity (Zafar et al., 2024).A seven-point Likert scale was used to rate each research issue, with 1 denoting strongly disagreed and 7 denoting strongly agreed.Three sections made up the questionnaire: the first asked about respondents' demographic profile using a nominal scale; the second asked about their prior experiences using social commerce applications; and the third section addressed the four constructs that this study looked at: intrusiveness, privacy perceived usefulness of AIassistance tools and behavioural intelligence.
Three professionals and academic specialists assessed each study item in relation to the research setting (i.e.social value) to proofread and validate the questionnaire.Gender and prior social commerce application usage were used in this study as behavioural intention control factors.The surveys were translated from English text into Arabic and Afghan using a back-translation technique to accommodate possible participants who spoke Arabic, Syrian, Saudi or Afghan as their official language.After extending an invitation to seven emarketing academics via email, the translated and original texts were shown to four of them to assess the degree of consistency between the two texts, the comprehensibility and clarity of the translated content, and the elimination of typos and grammatical errors.As a result, they highlighted the content's readability, went over each item and used it in the AI context.To establish item validity and reliability, this paper also conducted a pilot test with 50 responses.Findings indicate that the range of internal consistency reliability, or Cronbach's alphas, was within acceptable bounds, ranging from 0.881 to 0.907.Pilot studies can accept reliability results of 0.60 or higher (Pearson et al., 2020).According to Hair et al. (2019aHair et al. ( , 2019b)), numerous academics concur that Cronbach's alpha values should fall between 0.70 and 0.90.

Sampling and procedure
According to DATAREPORTAL's (2023) Saudi Arabia Digitization 2023 report, the research population totalled 29.10 million social media users.Because of the multiplicity of populations in Saudi Arabia, the most census social media users were chosen as the unit of analysis.As a result, such users were chosen, whether they were Sudanese, Syrians, Afghans or Saudi residents; hence, a purposeful sample strategy was used.This method does not require a preset sample size or theoretical foundations (Etikan et al., 2016), but rather identifies and selects individuals or groups who are aware of and talented about the issue of interest.
According to Sarker and Al-Muaalemi (2022), some advantages of this technique are as follows: does not require randomisation; assists researchers in creating a precise sample at a cheap cost; is related to low error rates; and relies on the researcher's discretion when selecting study units (e.g.participants, instances or organisations).
The fundamental disadvantage of this sample strategy, on the other hand, is that bias is possible because participants are aware they are participating in the study.There is insufficient capacity to generalise existing findings.
In a pilot study, 80 social media users were invited.These individuals were contacted by sending a pre-made questionnaire link through Google Forms to Facebook groups that promoted AI-powered social commerce solutions in Saudi Arabia, including Buffer, Ocoya, Taplio and Publer.It was made clear to the participants that participation was entirely optional.Fifty participants replied to their invitation within a week, offering helpful recommendations such as shortening the item and explicitly rephrasing "PVC2", one of the privacy issues.In this sense, within two days of sending out their invitation, researchers gave $5 charging cards for smartphones to those who replied.Their phone numbers were obtained by making them a required field on the questionnaire that was sent to them along with their demographic data.The final questionnaire was refined in light of these considerable efforts to get ready to gather the primary data set.

Data gathering
Users who had previously used social networking sites during the pre-test provided the necessary information via the Google Forms platform.Using a pre-made link on this platform that was sent over Messenger, 900 respondents were contacted between April and May of 2023.Therefore, a total of 509 social media users in KSA provided completed responses, yielding a response rate of 56.56%.21 replies showed outlier values that were more than 10% of the allowed limit, according to data screening with SPSS v.28.488 sample responses were the final sample size as a result.Since Krejcie and Morgan (1970) found that Fostering social media user intentions a sample size of 384 is needed if the population is more than one million, they concluded that this size was adequate.

Analysis strategy
Using SmartPLS v.4, the structural equation modelling (SEM) method was used to evaluate hypotheses following Hair et al.'s (2019a) recommendations.A more thorough approach to data analysis is required due to the different correlations that exist between the targeted components (Raza et al., 2023;Sarstedt et al., 2022;Waqas et al., 2023).For examining possible connections between several constructs with mediation effects in marketing, the partial least squares (PLS) approach is a great substitute (Hair et al., 2019b;Shahzalal and Elgammal, 2023).Hair et al. (2020) state that the PLS technique has the potential to investigate the complexities of formative and reflective models.
PLS-SEM is suitable due to its accuracy with complex models with a limited sample size (Shoukat et al., 2023b).Work-life quality was added as an external construct in the markercorrelation test (MCT) conducted to examine common method bias (CMB) concerns.The current data set showed no discernible bias, demonstrating that the model without MCT was superior to that with MCT.Given these explanations, CMB is not a significant concern in this paper.

Respondent profile
Table 1 demonstrates that 54.8% of respondents were males, and 47.4% of them were in the 21-30 age range.In total, 64.3% of respondents have a bachelor's degree.A total of 63.2% of respondents had sufficient expertise with AI-powered applications from three to less than five years ago because of their prior use of digital platforms.Moreover, the following nations made up the participants: Syria accounts for 38.7%, Saudi Arabia for 27.5%, Sudan for 21.3% and Afghanistan for 12.5%.

Outer model assessment
In the second stage, internal consistency and reliability are assessed; Jöreskog's composite reliability (CR) is most commonly used for this purpose.According to Hair et al. (2020), higher values signify improved reliability.According to Hair et al. (2019a), CR values of 0.95 and higher are troublesome because they imply that items are redundant.For instance, CR levels of 0.70 to 0.90 range from acceptable to good.The results of Table 2 show that the CR values ranged from 0.901 to 0.918.Besides, Cronbach's alpha follows the same assumptions as composite reliability but produces less precise results.However, constructs' reliability is thought to be somewhere between these two extremes.Table 2 data show that alpha values ranged from 0.871 to 0.907, indicating strong internal consistency and reliability.
The third stage focuses on the convergent validity of each construct measure using the average variance extracted (AVE), which means that each indicator loading on a construct must be squared before calculating the average value.An appropriate AVE of 0.50 or above suggests that the concept explains at least 50% of the variation in its items.Table 2 reveals that AVE values exceeded 0.50 (range from 0.693 to 0.750).This verifies the model's remarkable convergent validity.
In the fourth stage, the variance inflation factor (VIF) values were extracted to determine item collinearity, which confirmed that values greater than 3.3 (Hair et al., 2020) implies that there are no variations in gathered responses on any given item, whereas values less than this limit indicate a lack of multicollinearity for calculated construct items.As a result, Table 2's findings demonstrate that VIF values were between 1.486 and 2.467, showing that the current data set is not multicollinear or biased.

Fostering social media user intentions
The discriminant validity, or last step, can be conducted to assess a construct's empirical separation from other constructs in the structural model (Hair et al., 2019b).The heterotraitmonotrait (HTMT) ratio is calculated as the geometric mean of the average correlations for items measuring the same construct divided by the mean value of item correlations across constructs.Sarstedt et al. (2022) advocate a lower, more cautious threshold value (HTMT less than 0.85) when conceptions are semantically more diverse.Table 3 demonstrates that the HTMT values did not surpass the 0.85 criterion, confirming that the outer model has sufficient discriminant validity.

Inner model assessment
After the outer model was determined to be appropriate, the inner model was assessed using a PLS-bootstrapping technique (Schuberth et al., 2023).Common evaluation metrics to be considered include coefficient of determination (R 2 ), cross-validated redundancy metric (Q 2 ) and statistical significance of path coefficients (Hair et al., 2020).VIF values greater than 5 indicate that there might be issues with collinearity among the predictor constructs (Hair et al., 2019b).Ideally, the VIF values are less than 3.3 (Hair et al., 2020).Results shown in Table 3 verify that the VIF values varied from 1.000 to 1.843, suggesting that construct collinearity is absent from the current data set.
The first step is to look at the R 2 value, which measures how much variation each endogenous element explains and how well the model explains the current data set (Hair et al., 2020).Hair et al. (2019a) classify R 2 values of 0.75, 0.50 and 0.25 as strong, moderate and weak, respectively.As a result, Table 4 demonstrates that the R 2 value explained 53.9% of the variance in perceived usefulness, demonstrating moderate explanatory power.Otherwise, R 2 explained 40.6% of the variance in behavioural intention, demonstrating adequate explanatory power (Hair et al., 2019b).Calculating the Q 2 value is another technique to assess the model's prediction ability (Hair et al., 2019a).Q 2 values greater than zero indicate significant predictive relevance.Table 4 results demonstrate that Q 2 of perceived usefulness was 0.446 and Q 2 of behavioural intention was 0.375, validating the current structural model's medium predictive significance (Schuberth et al., 2023).
In the final stage, two criteria are used to evaluate model fit, as follows: standardised root mean square residual (SRMR) and normalised fit index (NFI).According to Schuberth et al. (2023), SRMR is the difference between the observed correlation and the correlation matrix.Hence, SRMR allows the model fit criterion to be evaluated by calculating the average size of the discrepancies between observed and anticipated correlations.To avoid model misspecification, SRMR values less than 0.10 or 0.08 are considered satisfactory (Hair et al., 2019a).Table 4 shows that the SRMR value was less than 0.08 (SRMR ¼ 0.048), validating the current model's goodness of fit.Otherwise, NFI computes the suggested model's Chi 2 value and compares it to an appropriate benchmark (Schuberth et al., 2023).NFI values above 0.90 often indicate a satisfactory match (Hair et al., 2019b).As a result, Table 4 reveals that the NFI value exceeded 0.90 (NFI ¼ 0.934).

Discussion
This article explored the psychological aspects (privacy and intrusiveness issues) that influence Saudi Arabian customers' perceptions of AI-assistance tools' usefulness on social commerce platforms, as well as their behavioural intent to embrace them.The findings demonstrate an appreciation for AI's ability to persuade users to give personal information, which is the driving force behind AI-powered applications that will improve organisations' access to a large amount of knowledge that computational capacity can analyse.According to Kronemann et al. (2023), research into how AI affects user behaviour is still in its infancy.This research is an early attempt to investigate exposures to user privacy information in the context of artificial intelligence.
The current model has only moderate support when compared to TPB. Results demonstrate that, in line with H1, users' behavioural intentions were positively impacted by perceived usefulness.Findings corroborate those of Selem et al. (2023a), who discovered that attitudes towards using social commerce applications and purchase decisions may are influenced by perceived usefulness.This implies that an AI-powered tools' effectiveness influences customers' evaluation of its benefits during the purchase process.
The results of H2a show that behavioural intention is not impacted by intrusiveness concerns in the context of the current investigation; this conclusion indicates a negative link Fostering social media user intentions between behavioural intention and intrusiveness concerns.This result can also be supported by Saudi research communities.Customers are aware that the Saudi Arabian business sector is characterised by the development of new technologies (i.e.cybersecurity), using contemporary hardware and software to protect business and customer data from any intrusion, disruption, alteration or illicit exploitation.This is because few previous studies have examined the relationship.
Findings show that the significance of intrusiveness concerns in behavioural intention was not affected by the mediation effect of perceived usefulness (H2b is not supported).If not, consumers' intentions regarding their behaviour are negatively impacted by privacy concerns.Due to their concerns about making purchases using social commerce platforms, high-privacy consumers' online social lives are hampered (Higueras-Castillo et al., 2023).This outcome is associated with a higher frequency of use of AI-powered applications by men.Women are more private than men are, especially in Arab nations where customs and conventions protect women's privacy and their apprehension about prying eyes and outside access.
In H3b, the presence of perceived usefulness will exacerbate the negative nexus of privacy concerns with user intentions, resulting in partial mediation.Thus, regardless of the benefits provided to customers, privacy is an important consideration while purchasing.To boost consumer and business trust, industry leaders in online commerce must provide their systems with the highest online security requirements.

Contributions
Using perceived usefulness as a mediator, this study is theoretically one of the few that investigates the impact of psychological aspects (such as privacy and intrusiveness concerns) on users' purchasing intentions via social media.By expanding user privacy research and bolstering the TPB framework with AI-powered social commerce platforms for Saudi Arabian consumers, this study considerably advances the theory.The current model examines privacy and intrusiveness issues that affect users' adoption of social commerce platforms with AI capabilities (Table 5).The essential components of TPBbehavioural intentions and perceived usefulnesswere combined to meet the study's objectives.
To the authors' knowledge, much research has focused on social commerce in general (Abed, 2020;Selem et al., 2023d) rather than the AI context, making this study one of the first to investigate AI-powered social commerce platform adoption.Because there has been little research on the variables leading users to be concerned about privacy and intrusion in the Saudi context, this paper proposes a theory about what causes users to trust social media.The article increases understanding of how privacy cues and fear of intrusion influence purchasing decisions.The study also shows that customers' perceived advantage, which is the primary purchasing motivation, affects their privacy and intrusion worries.
From a practical standpoint, this study offers practitioners valuable information to support the development and execution of strategic initiatives aimed at promoting customers' online shopping behaviours (Table 5).We can use the operational recommendations from this study to encourage customers to make secure online purchases.Managers and retailers must recognise that consumer behaviour has shifted in terms of the context of perceived usefulness.They ought to make an effort to improve the purchasing experiences of customers by bringing in technological innovations and increasing the social media options available to shops.
Practitioners should select e-commerce technologies that are fully integrated with Facebook and Instagram.Integrating chatbots and AI-powered customer service can greatly improve consumer interactions by responding rapidly to inquiries and recommending products.It is vital to offer clients secure and user-friendly payment methods.Investing in solid cybersecurity measures is crucial for securing customer data.AI for influencer marketing can help find influencers whose values and brand alignment are relevant to the target audience (Shoukat et al., 2023c).As a result, buyers will be less concerned about their privacy and more confident, boosting the possibility of purchasing via social media.
The current findings can assist Saudi Arabian policymakers in better positioning their strategies for fostering customer confidence while encouraging social media utilisation of amenities, given that the country is still evolving and that its customers lack confidence and are reluctant to embrace social media for e-commerce services.By enhancing platform attributes including data integrity, simplicity, safety and secrecy, social media managers can increase user trust (see Table 5).Nonetheless, other factors, such as privacy concerns, are just as important as trust in influencing purchase decisions over social media.To enhance the efficiency of privacy and security requirements, it is also advisable to avoid lengthy statements and technical jargon that the general public finds confusing.Customers' impressions of privacy and security standards may improve if they see these policies in video format.

Limitations and future research directions
Initially, this study focuses on a single country: Saudi Arabia.These findings cannot be generalised because of cultural differences across countries.Future research should take into account other countries' cultures when adopting findings and European and Arab countries should be picked for a comparison study.The only variables under control in this article were the participants' ages and social media experiences.Future studies can make use of additional demographic parameters, such as education.The findings may be less generalisable because the data analysis was based on a representative sample of social commerce application users.Future researchers could use a multi-tiered approach to user types (consumers, customers and repeaters).
Furthermore, numerous elements were explored to determine the causes of customers' social media preferences for products and services.TPB was used to construct the theoretical model.Future studies may include other explanatory factors and theories to -Enhance consumers' buying experiences by introducing technical advancements and expanding social media alternatives for retailers' offerings Users' favorable opinions of social commerce platforms can be shaped by the perceived usefulness of these AI-powered platforms and the advantages they offer -Integrating chatbots and AI-powered customer service can significantly improve customer interactions by responding quickly to inquiries and making product recommendations If e-business can provide comfort, users will appreciate it, and their intentions to adopt it will be strengthened -Managers must provide their services on electronic platforms in a way that covers all customer inquiries and reflects the benefits of the products provided Female users in Arab countries tend to value privacy and are reluctant to divulge personal information -It should advance the understanding of how privacy cues and fear of intrusion impact purchase intentions through secure, technology-mediated channels -It is critical to provide customers with secure and user-friendly payment options.Investing in strong cybersecurity measures is critical to protecting customer data Fostering social media user intentions develop more comprehensive models of technological acceptance.This work was restricted to investigating the effects of privacy and intrusion concerns on social media user intentions.Future research could look into other characteristics such as purchasing decisions, ability to pay and eWOM.A survey was used to examine all of the study's constructs simultaneously.As a result, future research should test the suggested model using various techniques, such as experimental or quasi-experimental methods.
Finally, future research can test the moderation function of religiosity on purchasing behaviour in the virtual arena of online shopping.Furthermore, the prior study did not address the cross-cultural issue.Thus, the suggested model can be tested in diverse cultural situations, allowing for a comparative comparison of emerging and established economies in future research (Batabyal et al., 2023;Sinha et al., 2023).

Conclusion
AI-powered social commerce platforms are a current business trend that involves interactive purchasing and selling platforms.It enabled direct user connection and gave businesses honest customer feedback.The perceived usefulness of these AI-powered platforms might influence consumers' positive sentiments towards social commerce platforms based on the benefits they provide.Users appreciate comfort, for example, with tangible rewards (e.g.difficult-to-purchase brands) and moral benefits (i.e. the convenience of online purchasing) (Elgammal et al., 2022); if e-business can provide these, user adoption intentions will improve.
Furthermore, the findings suggest that female users prefer privacy and are afraid to share their data, particularly in Arab nations.Despite its faults, this article provides a detailed explanation of users' adoption behaviour via AI-powered social commerce platforms, as well as the factors that favor/hinder adoption.This is the first paper to look at how privacy issues influence behavioural intentions on AI-powered commerce platforms (Table 5).