1. Introduction

The emergence of generative artificial intelligence (GenAI) applications, such as ChatGPT, Gemini, and Claude, is transforming how individuals access online information, including political information and news. Given the increasing integration of GenAI into online search, messenger, and other services, along with its unique ability to serve as a hyper-personalized universal information source with dialogue-style interactions, these technologies could significantly lower barriers to news consumption. This may alter news consumption patterns in a way similar to the disruptions caused by the rise of search engine-based news aggregators and social media; that is, those who are already very interested in news may benefit the most from the emergence of a new technology (Kümpel, 2020; Merten et al., 2022; Prior, 2007; Strömbäck et al., 2022). However, individuals who traditionally engage less with political news, such as those with lower political interest, lower trust in media, or a tendency to avoid news, could begin using GenAI for political information, potentially inspired by the proven helpfulness of the technology in other areas of life and its adaptability to one’s interests and literacy level.

Fears exist that using GenAI for news and political information could lead to the replacement of traditional news sources. Two important risks are connected to this. First, GenAI is associated with factuality issues that are often difficult for laypeople to detect (Augenstein et al., 2024). Although this is an important issue in itself, the problem is exacerbated when Gen­AI-based news consumption becomes the reference and replaces journalistic news sources, which then no longer serve as a layer of verified, reliable information. This may contribute to people more easily accepting and sharing misinformation. Second, online news providers are fundamentally dependent on ad-based revenue. GenAI poses a serious risk to this business model because it allows users to access and consume the required information, frequently derived from journalistic sources, without directly engaging with original content providers or news platforms. These actions undermine the ad impressions and clicks on which these outlets rely (Perez, 2025; Reisner, 2025).

These speculations about the opportunities and risks associated with GenAI in the democratically consequential field of political information and news warrant empirical monitoring, particularly of how the adoption and usage of this technology, and its effects, manifest in practice. In light of this, the present paper explores the extent to which GenAI replaces or substitutes news consumption among people with low news consumption and people who intentionally avoid news. The relevance of this research is underscored by Google’s 2024 shift from traditional search results pages to AI-generated answers resembling GenAI interfaces (Reid, 2024). Given Google’s dominant position in Internet use worldwide and the fact that search engines have traditionally been regarded as reliable and trustworthy sources of information, including political content (Kelly & Fu, 2007; Puschmann, 2019), it is essential to understand how users adopt GenAI to acquire online information.

This study addresses this gap by examining how GenAI is used for political information and news. Drawing on a conceptual discussion of GenAI’s affordances for information and news use, it investigates whether GenAI usage aligns with existing news consumption patterns or instead fosters engagement among groups that have traditionally been less involved in news consumption. Using a representative survey conducted in Germany before the 2024 European Parliament elections, we analyze the extent to which GenAI alters patterns of political news consumption and explore its potential implications for the democratic information landscape.

Our findings show that, so far, GenAI is rarely used for political information and news. Furthermore, those with higher political interest, higher trust in news, and richer news media diets engage more with GenAI for political information and news. Thus, familiar patterns of digital news consumption likely also apply to GenAI. However, some associations are not consistent across all analyses, including the effect of education, while others diverge from established patterns, most notably the negative association between the sense of duty to stay informed and the use of GenAI for political information and news. The implications of these findings are discussed and highlight the need for additional research.

2. Use of GenAI for Political Information

GenAI was made accessible to a broad audience with the launch of ChatGPT by OpenAI in November 2022 and immediately attracted immense attention. By 2026, millions of people worldwide had tried the technology, including 34% of the US population and about 44% of the German population (Chatterji et al., 2025; Reiss et al., 2025; Sidoti & McClain, 2025). Adoption rates continue to rise steeply, suggesting that a majority of the population may become familiar with the technology within the next few years. In addition, the first empirical studies found that a younger age and higher education are positively associated with using ChatGPT, the most popular text-based GenAI application (Kacperski et al., 2024; Reiss et al., 2025).

As concerns the uptake of GenAI in the context of political communication and news, research has mainly focused on the adoption of GenAI in journalism and newsrooms (Dodds et al., 2025; Nishal & Diakopoulos, 2024) and how news audiences perceive AI-generated news (Morosoli et al., 2025; Toff & Simon, 2024; Wang & Huang, 2024). On the user side of GenAI use, studies have so far primarily examined science- and health-related information contexts (Ayre et al., 2025; Bautista et al., 2025; Greussing et al., 2025). Only limited and high-level descriptive insights exist regarding GenAI use for news, suggesting that a small fraction of the population uses the technology for news consumption (Lipka & Eddy, 2025; Newman et al., 2025; Reiss et al., 2025). To gain a better understanding of how GenAI is used for the consumption of political information and news, we approach this issue in an exploratory yet systematic manner. Our analysis is primarily anchored in the contemporary discussion of news consumption and news avoidance, and we derive assumptions about the use of GenAI for news from what we identify as the key affordances of this technology in terms of information access.

3. Affordances of GenAI and Their Potential Impact on News Consumption

The concept of affordance, proposed by Gibson (2014), refers to the range of actions that an environment makes possible. Applied to technology, it captures the dynamic relationship between users and media and results from the interaction of distinct characteristics of the technology, users’ perceptions, and the broader social context in which the technology is embedded. Emerging communication technologies can thus create novel affordances and expand the range of actions available to end users (Evans et al., 2017).

Three key affordances of GenAI, namely, hyper-personalization, dialogue-style interaction, and universality, are central to understanding its potential effects on political information consumption.

1. Hyper-Personalization: Unlike traditional algorithmic personalization, which selects content from a predefined set based on user behavior (Just & Latzer, 2017), GenAI generates tailored responses based on user input. This could lower barriers to news engagement by mitigating common reasons for news avoidance, such as negative emotional effects, information overload, and distrust in traditional media (Skovsgaard & Andersen, 2020). However, hyper-personalization also introduces risks such as biases from the training data, user input, or GenAI’s tendency to mimic users’ sentiment (Blassnig et al., 2023; Ferrara, 2023; Simmons, 2022), as well as various factuality challenges (Augenstein et al., 2024).
2. Dialogue-Style Interaction: GenAI facilitates intuitive and interactive engagement, allowing users to ask follow-up questions and explore topics in depth (Radlinski & Craswell, 2017). This could enhance users’ understanding and benefit individuals who find traditional news formats overwhelming or difficult to engage with (Luskin, 1990; Palmer et al., 2020). Although human–AI interactions do not replicate interpersonal discussions, which are important to some in the context of political information (Peters et al., 2022; Reiss et al., 2021), they may serve as an alternative source for people who struggle with conventional news sources.
3. Universality: GenAI consolidates diverse types of data, making it a convenient one-stop shop for information, including news, educational content, and entertainment. This contrasts with traditional online news sources and information in general, which require users to navigate multiple websites or apps.

These affordances present both opportunities and risks in terms of disrupting current patterns of news consumption, and they have distinct implications at the level of individuals. The subjective value and effort associated with these affordances therefore serve as an analytical lens for examining the potential effects of GenAI on news consumption. Consequently, in the following section, we first discuss the potential of GenAI and its affordances to attract audiences with low news consumption and intentional news avoiders. We then explore concerns that GenAI might crowd out traditional journalistic news sources by replacing them or diminishing their role in the information ecosystem.

4. GenAI’s Potential for Disrupting Inequality in News Consumption

Research has consistently shown that as the media environment has shifted toward greater choice, inequalities in news consumption have deepened (Aalberg et al., 2013; Dahlgren, 2019; Karlsen et al., 2020; Kümpel, 2020; Merten et al., 2022; Prior, 2007; Strömbäck et al., 2013; Thorson, 2020). Although the affordances of social media, particularly its potential for incidental exposure to news, may have somewhat mitigated these developments, digitalization and a high-choice media environment have nonetheless widened gaps in political knowledge and participation (Damstra et al., 2023; Delli Carpini, 2000; Leeper, 2020; Lind & Boomgaarden, 2019; Prior, 2005; Strömbäck et al., 2022). Alongside growing disparities between groups, the number of people who sometimes or often avoid the news has increased globally, rising by 38 % since 2017 (Newman et al., 2025). Both the general decline and inequalities in news consumption and political engagement across social groups may threaten democratic systems, which rely on an informed citizenry and equal representation and participation by all citizens (Delli Carpini, 2000; Habermas, 1989).

However, patterns of news consumption, engagement, and avoidance vary significantly across different population groups, as do the underlying reasons for these differences and their perceived normative implications. Hence, to systematically assess the disruptive potential of GenAI for different groups, we follow a recent suggestion by Betakova et al. (2025) and analytically differentiate between the dimensions of news consumption and intentional news avoidance. Whereas the former is an observational, external criterion, the latter describes a self-identified behavior related to subjective intention and motives. The two dimensions complement each other.

1. Low News Consumption and GenAI

No or low news consumption reflects a general detachment from media and politics and is rooted in personal characteristics such as individual preferences, low political interest, and low media and political trust (Betakova et al., 2025; Lecheler & de Vreese, 2017; Prior, 2005; Skovsgaard et al., 2016). These factors tend to be stable over time (Prior, 2010). Often, these people have a higher preference for entertainment and favor such content over news (Prior, 2005). While there is little consensus on what level of news consumption qualifies as low or non-use (Reiss, 2023), it is established that low or non-consumption is related to undesirable normative outcomes, such as low political knowledge, engagement, and civic participation (Damstra et al., 2023; Edgerly et al., 2018; Lecheler & de Vreese, 2017). Furthermore, a well-informed citizenry is more resilient to mis- and disinformation (Altay et al., 2023; Humprecht et al., 2020; Mont’Alverne et al., 2024).

The implications for GenAI use in this context are nuanced and hinge on how GenAI’s affordances influence the subjective value and effort in news use. The hyper-personalization of GenAI may lower barriers to understanding and increase the perceived relevance of news, while its dialogue-style communication can foster a more active and engaged user role by enabling explanations of content and clarifying follow-up questions. Together, these affordances may enhance accessibility, interest, and information processing (Gao et al., 2024), making GenAI particularly attractive for groups that have traditionally been less engaged with news, such as women, younger people, and individuals with low education levels. An uptake by these groups could help reduce inequalities in news engagement.

Further, low traditional news consumption is often associated with low media trust (Betakova et al., 2025). However, GenAI may be seen as a more trustworthy and non-mainstream news alternative, and therefore as more attractive, particularly because AI and machines are often perceived as more neutral and objective than humans (Araujo et al., 2020; Logg et al., 2019; Xavier, 2025). The universality of GenAI, with users already familiar with the technology from other everyday contexts, may further lower barriers to use and increase its perceived usefulness and acceptability as a news source for some users.

At the same time, GenAI’s hyper-personalization and dialogue-style communication require additional effort due to the need for active user input. Given that substantial segments of the population abstain from actively seeking news and rely instead on passive and incidental exposure, such as on social media feeds (Boczkowski et al., 2018; Gil de Zúñiga et al., 2020; C. S. Park & Kaye, 2020; Searles & Feezell, 2023), the requirement of active user input makes it unlikely that individuals with little or no interest in politics and news will proactively request such information, as this requires more effort than visiting and skimming a news website. Moreover, prompting GenAI for news presupposes a certain level of prior knowledge, which may be even more prohibitive for many users.

Hence, GenAI could likely contribute to a further widening of the gap between news enthusiasts, who benefit from a new opportunity for in-depth, tailored engagement and discussion around news, and people with little interest in political information, for whom the required effort may outweigh the perceived value. This would undermine expectations that GenAI could have a transformative impact on political information.

2. Intentional News Avoidance and GenAI

In addition to (low) news consumption, the behavior of intentional news avoidance can be understood as the active decision to take a break from emotionally burdensome news (Betakova et al., 2025). Intentional news avoidance can be situational or temporary, and Betakova et al. (2025) even suggest that it is a by-product of news use. Reasons to avoid news can relate to low trust in news (Strömbäck et al., 2020; Toff & Kalogeropoulos, 2020), as discussed above for low news consumption. Further reasons are a perceived news overload (Holton & Chyi, 2012; Schmitt et al., 2018; Song et al., 2017) or the negative impact of news on subjective well-being (Betakova et al., 2025; Boukes & Vliegenthart, 2017). Using GenAI for news may alleviate both factors: hyper-personalization by GenAI allows users to have a highly selective news experience that can be limited to a preferred topic or a summary of news items. Similarly, issues that are subjectively burdensome or of no interest (e.g., war or conflicts) can be excluded, and news can be presented in a positive or constructive style, which is often suggested as a potential solution to news avoidance (Overgaard, 2023; Palmer & Edgerly, 2024). Given that intentional news avoidance is high both globally and in Germany (Behre et al., 2025; Newman et al., 2025), GenAI could become a tool for a more selective, less overwhelming, and, thus, more emotionally sustainable and subjectively valuable form of news consumption. Furthermore, since intentional news avoidance is less predominantly driven by a lack of political interest (Goyanes et al., 2023), the effort required for active user input in Gen­AI is less likely to be a limiting or prohibitive factor. As a result, GenAI may prove particularly transformative for intentional news avoiders. However, most people with high intentional news avoidance have an average or high news consumption (Betakova et al., 2025; Palmer et al., 2023), so this dimension is normatively less relevant at a societal level.

Finally, feeling a duty to be informed is positively associated with news consumption and negatively with news avoidance (McCombs & Poindexter, 1983; Palmer & Toff, 2020; Toff & Nielsen, 2022; Trilling & Schoenbach, 2013). As individuals weigh the time, mental effort, and emotional energy required to follow the news (Palmer & Toff, 2020) against the value they get from fulfilling their intrinsic informational duty, the easy access to news and hyper-personalized, subjectively more relevant information offered by GenAI may significantly improve this trade-off. Hence, for people with a strong duty to stay informed, GenAI may alter the subjective value–effort balance, offering some transformative potential in the context of political information and news use, at least for some individuals.

Given the lack of a clear overall picture of the potential of GenAI to transform news consumption and disrupt existing information inequalities, we pose the following exploratory research question:

RQ1: To what extent do patterns of low news consumption and intentional news avoidance relate to the use of GenAI for news and political information?

5. GenAI’s Potential in Disrupting the Established News Ecosystem

Alongside the chance that GenAI will attract individuals who are less connected to news and political information, there are also fears that the use of GenAI for such purposes could replace traditional news sources. The justification for these concerns lies in the individual appeal and value of GenAI, which can be attributed to the affordances of the technology. First, unlike traditional news formats, such as websites or television, GenAI’s hyper-personalization delivers news with greater subjective relevance while reducing the time and effort required to access it (Monzer et al., 2020; Wu et al., 2022). Second, dialogue-based interaction with GenAI enhances user convenience by allowing individuals to express their information needs in natural language and enables them to avoid potentially overwhelming news homepages (Radlinski & Craswell, 2017). Additionally, it allows for follow-up questions and deeper engagement with the topic. Third, since users are often already and increasingly engaged in conversation with the technology for multiple purposes (Reiss et al., 2025; Skjuve et al., 2024), there is no need to switch platforms or contexts; GenAI can provide a wide range of information, including news, in one place, further adding to the user’s convenience. From the user’s perspective, these affordances of GenAI are intertwined and may contribute to a feeling of being fully informed, reducing the perceived need to consult additional sources, similar to the “news-finds-me” perception observed in the context of news consumption via social media (Gil de Zúñiga & Diehl, 2019). However, in contrast to exposure to news on social media, which is often rather passive, GenAI requires active user input. For some users, this may make visiting a news website and engaging in passive scrolling comparatively more attractive than the effort required to prompt GenAI for similar information. This unresolved issue leads to the second research question:

RQ2: To what extent is the use of GenAI for news and political information associated with the frequency of journalistic news source consumption?

6. Methods
   1. Survey and Participants

For a nuanced investigation of the research questions, we relied on an online survey conducted in Germany in the two weeks prior to the 2024 European elections on June 9th. The survey sample included 1,461 individuals1 aged 16 years and above and was representative of the online population in Germany in terms of age, gender, education, and state of residence. Participants were recruited by the German panel provider GIM through both offline and online methods in roughly equal proportions. Participants received a small financial compensation for their participation and provided informed consent before starting the survey.2 The median survey time was 24 minutes. Weighted means and proportions are reported to represent population estimates, while raw counts show the actual sample sizes. The questionnaire is attached in the Online Appendix.3

A total of 43.8 percent of participants (n = 608) reported using GenAI. Compared with the full sample (n = 1,461), GenAI users featured a lower share of women (45 % vs. 49.5 %), were younger (median age 30–39 years vs. 40–49 years), and had a higher proportion of individuals with higher education (53.5 % vs. 39.1 %). Table 5 in the Online Appendix provides more details on the sample.

2. Variables and Operationalization

News consumption, intentional news avoidance, and GenAI use for political information and news are the central variables for answering RQ1 of this study. We measured news consumption as the mean consumption of a list of 15 news sources, including six online and offline journalistic sources (e.g., news websites or TV news) and nine non-journalistic sources (e.g., social media or family and friends). For each news source, participants indicated the frequency of their exposure during a week on a six-point scale of 1 = “never” to 6 = “multiple times a day” (M = 2.29, SD = 0.82). Intentional news avoidance was assessed with a single item: “Have you been actively trying to avoid news lately?” Responses were given on a five-point scale of 1 = “never” to 5 = “very often” (M = 2.32, SD = 1.32). Following Betakova et al. (2025), we operationalized low news consumption and high intentional news avoidance as a binary variable, identifying participants whose overall news consumption was more than one standard deviation below the mean (i.e., cutoff of 1.47) and one standard deviation above the mean (i.e., 3.64), respectively.

Regarding the use of GenAI for news, three operationalizations were used to capture more nuance. The three outcome variables are: using GenAI (1) for general political information, (2) as a substitute for news, and (3) for electoral information on the 2024 European elections. These questions were only asked of the 43.8 percent of participants who reported using Gen­AI applications, and answers were given on a five-point scale ranging from 1 = “never” to 5 = “very often” (M = 1.59, SD = 1.32).

Furthermore, to answer RQ2, using GenAI for news was coded as a binary variable. High GenAI news usage identified participants whose mean usage of GenAI for news (across the three variables, Cronbach’s α = 0.87) was one standard deviation above the mean (i.e., cutoff of 2.48). Consumption of journalistic news sources represents the mean consumption of a list of six journalistic news sources (TV, print, radio, podcasts, websites/apps of traditional news outlets, and websites/apps of online news outlets), and the number of journalistic news sources is the count of news sources with a consumption higher than “never.”

Additionally, political interest was measured on a five-point scale ranging from 1 = “not at all” to 5 = “very interested.” News media trust was measured as the average trust across six German outlets (ARD, ZDF, Spiegel, Bild, Web.de, Freie Welt) on a five-point scale from 1 = “I do not trust them at all” to 5 = “I fully trust them” (α = 0.76). Sense of duty to stay informed is the average of four items (Gil de Zúñiga et al., 2017; McCombs & Poindexter, 1983) with answers on a five-point scale ranging from 1 = “I do not agree at all” to 5 = “I completely agree” (α = 0.75). A higher value implies a more pronounced sense that staying informed is important. Furthermore, since attitudes toward technology can influence its adoption (Edison & Geissler, 2003; Kai‐ming Au & Enderwick, 2000), subjective attitude toward GenAI was included as a control variable. It was measured as the composite mean of 11 items assessing various aspects of attitudes toward GenAI (loosely based on Frewer et al., 1998; Gillespie et al., 2021), rated on a five-point scale from 1 = “does not apply at all” to 5 = “applies completely,” with higher scores indicating more positive attitudes (α = 0.86). Finally, experience with GenAI may affect users’ sophistication and familiarity, potentially leading to broader use, including for political information and news. Therefore, the duration of participants’ GenAI use (in months) was included as a control. Gender, age, and education were also added as controls. Details on all the variables can be found in Table 5 in the Online Appendix.

3. Analysis

To answer the first research question, we conducted t-tests comparing the use of GenAI for news (1) between low and average/high news consumers and (2) between low/average and high intentional news avoiders. In addition, we performed multiple linear regressions to examine the influence of the independent variables on the use of GenAI for news, while controlling for relevant variables. The same approach was applied to address the second research question.

7. Results

Overall, the findings indicate that the participants rarely used GenAI for political information and news. For the most common variant, namely, using GenAI for general political information, 63.5 % of GenAI users reported never using the technology for this purpose (scale minimum 1), while only 1.4 % said they did so very often (scale maximum 5). The mean usage was 1.68 (see Table 1; for the full frequency distribution, see Table 6 in the Online Appendix).

The study aims to examine the relationship between patterns of low news consumption and intentional news avoidance and the use of GenAI for news and political information. The results regarding this first research question are displayed in Tables 1 and 2. Looking at the isolated relationship, individuals with low news consumption reported significantly lower use of all three operationalizations of GenAI use for political information and news than participants with at least average news consumption (p ≤ 0.001). There was no significant difference in the use of GenAI for news between high and low/average intentional news avoiders.

Table 1: Comparison of means of using GenAI for news for news consumers (1) and news avoiders (2)

		Using GenAI
	N	For general political information	As a substitute for news	For electoral information
All	608	1.68 (1.04)	1.56 (1.02)	1.52 (0.95)
(1) News consumption
Low	95	1.31 (0.68)	1.26 (0.71)	1.12 (0.33)
Average/high	513	1.75 (1.08)	1.62 (1.05)	1.59 (1.01)
		p < 0.001	p < 0.001	p < 0.001
(2) Intentional news avoidance
High	117	1.79 (1.16)	1.69 (1.14)	1.56 (0.98)
Average/low	491	1.66 (1.01)	1.53 (0.98)	1.51 (0.94)
		p = 0.286	p = 0.181	p = 0.615

Note: P-values from t-tests comparing the means of the two subgroups.

To check for spurious correlations and gain a deeper understanding of the relationship, we conducted multiple linear regressions (Table 2). In these models, the relationship between high news avoidance and the use of GenAI remained statistically insignificant. However, for low news consumption, the results are more nuanced. In the more complex model, the relationship between low news consumption and using GenAI was statistically insignificant for two of the three operationalizations of GenAI news use in the regression analyses. The only exception was the use of GenAI for electoral information, where low news consumption continued to have a significant negative effect.

In all three regression models in Table 2, low news consumption and high intentional news avoidance accounted for only a small proportion of the explained variance (see Online Appendix Table 7 for details). In Model 1, high intentional news avoidance made the largest contribution, explaining 2.6 % of the variance. In contrast, the largest partial contributions across all three models were observed for media trust (29.9 %), political interest (26.4 %), and sense of duty to stay informed (16.8 %).

Table 2: Linear regression analyses for using GenAI for (1) general political information, (2) as a substitute for news, and (3) for electoral information regarding the European elections

	Using GenAI
	For general political information			As a substitute for news			For electoral information
	(1)			(2)			(3)
Low news consumption	-0.129	(0.101)		-0.133	(0.103)		-0.166	(0.070)	*
High intentional news avoidance	0.213	(0.120)		0.112	(0.127)		0.101	(0.106)
Age	-0.062	(0.027)	*	0.013	(0.030)		-0.022	(0.025)
Female	0.014	(0.084)		-0.008	(0.082)		0.145	(0.075)
Education	-0.105	(0.054)		-0.130	(0.055)	*	-0.118	(0.049)	*
Political interest	0.272	(0.046)	***	0.187	(0.046)	***	0.226	(0.040)	***
Media trust	0.390	(0.059)	***	0.299	(0.067)	***	0.380	(0.057)	***
Sense of duty to stay informed	-0.251	(0.060)	***	-0.289	(0.061)	***	-0.245	(0.059)	***
Positive attitude toward GenAI	0.219	(0.065)	***	0.134	(0.066)	*	0.165	(0.062)	*
Experience with GenAI	0.005	(0.003)		0.007	(0.004)		0.003	(0.003)
Constant	0.101	(0.383)		0.825	(0.380)	*	0.200	(0.362)
Observations	591			591			591
R2	0.216			0.153			0.196
Adjusted R2	0.201			0.137			0.181

Note: Standard errors in parentheses. * p < 0.05; ** p < 0.01; *** p < 0.001.

To address the second research question, which concerns the extent to which the use of Gen­AI for news and political information is associated with the frequency of consumption of journalistic news sources, we followed a similar analytical procedure.

Table 3 presents the results of the means for the comparisons of the consumption of journalistic news and the number of journalistic news sources between individuals with low/average and high use of GenAI for news. In both cases, those with low/average use of GenAI for news reported significantly lower consumption of journalistic news and fewer journalistic news sources.

Table 3: Comparison of the use of journalistic news sources between low/average and high users of GenAI for news

	N	Consumption of journalistic news	Number of journalistic news sources
All	608	2.26 (1.02)	3.23 (1.80)
Using GenAI for news
Low/average	514	2.12 (0.92)	2.97 (1.71)
High	94	2.93 (1.19)	4.49 (1.70)
		p < 0.001	p < 0.001

Note: P-values from t-tests comparing the means of the two subgroups.

These findings remained robust when controlling for additional relevant variables in regression analyses and when operationalizing GenAI use for news as a continuous rather than binary variable (Table 4). This points to the opposite of a replacement effect of journalistic news by GenAI. Further robustness checks of the reported results are presented in the Online Appendix.

Table 4: Linear regression analyses for the use of journalistic news sources: GenAI for news as (a, c) a binary variable and (b, d) a continuous variable

	Consumption of journalistic news						Number of journalistic news sources
	(a)			(b)			(c)			(d)
High GenAI use for news (binary)	0.616	(0.114)	***				1.205	(0.193)	***
GenAI use for news (continuous)				0.312	(0.045)	***				0.601	(0.076)	***
Age	0.193	(0.022)	***	0.199	(0.022)	***	0.183	(0.041)	***	0.195	(0.041)	***
Female	-0.062	(0.070)		-0.067	(0.069)		-0.141	(0.135)		-0.150	(0.134)
Education	0.021	(0.044)		0.028	(0.044)		0.003	(0.088)		0.016	(0.088)
Political interest	0.220	(0.035)	***	0.199	(0.035)	***	0.370	(0.069)	***	0.331	(0.070)	***
Media trust	0.391	(0.048)	***	0.356	(0.048)	***	0.675	(0.069)	***	0.611	(0.103)	***
Sense of duty to stay informed	0.114	(0.046)	*	0.137	(0.046)	**	0.265	(0.086)	**	0.310	(0.085)	***
Positive attitude toward GenAI	-0.010	(0.052)		-0.024	(0.051)		-0.164	(0.095)		-0.189	(0.095)	*
Experience with GenAI	0.001	(0.002)		-0.000	(0.002)		0.001	(0.004)		-0.001	(0.004)
Constant	-0.826	(0.252)	**	-1.107	(0.256)	***	-1.264	(0.461)	**	-1.809	(0.456)
Observations	591			591			591			591
R2	0.404			0.418			0.345			0.361
Adjusted R2	0.393			0.408			0.334			0.350

Note: Standard errors in parentheses. * p < 0.05; ** p < 0.01; *** p < 0.001.

8. Discussion

The results indicate that GenAI is not yet frequently used for political information and news. Moreover, its use in this context mirrors traditional news consumption patterns instead of diverging from them. Depending on the inclusion of controls, low news consumption was either unrelated to or negatively associated with GenAI use for news. In other words, we found no indication that people with low news consumption turn to GenAI for political information. Instead, the opposite appears true: individuals with higher levels of news consumption are more likely to use GenAI for this purpose.

The assessment that GenAI seems to widen gaps between news enthusiasts and those not interested in politics is further supported by the strong positive effect of political interest on using GenAI for news. This finding is in line with the effects of online news media on the spread of political knowledge in a population (Lind & Boomgaarden, 2019). This means that, so far, the affordances of GenAI, such as hyper-personalized content, intuitive interaction, and universality, have not translated into a perceived subjective value of engaging in news consumption through this technology. People’s disinterest in news, combined with the additional effort associated with necessary active inquiry when using GenAI for news and political information, arguably outweighs the potential benefits of using GenAI in this context. Furthermore, we did not observe consistently that younger people, females, or people with lower education – groups that typically consume less news – were positively associated with using GenAI for news. Finally, the strong positive association between media trust and using GenAI for news contradicts the idea that people with low media trust turn to GenAI for news. In summary, our findings provide little support for the notion that GenAI will disrupt current patterns of news consumption, and given the non-significant effect of experience with GenAI, such a change appears unlikely even when GenAI becomes used more heavily in the future.

One exception to the overall picture concerns the results on the sense of informational duty, which also accounts for a large share of the explained variance in the regression models. Prior research has shown that this feeling is positively associated with news consumption and negatively with intentional news avoidance (McCombs & Poindexter, 1983; Palmer & Toff, 2020; Toff & Nielsen, 2022; Trilling & Schoenbach, 2013). Our data confirm these associations. However, they also reveal a negative relationship with the use of GenAI for political information and news, similar to the relationship with consuming news via online messengers and smart speakers. Likewise, we identified a strong conditional effect of political interest on the sense of duty to stay informed. Once political interest is accounted for, the remaining component of informational duty seems to primarily reflect aspects that are negatively associated with news use (e.g., normative pressure, obligation, or feelings of burden or guilt). This finding aligns with recent arguments that political interest is increasingly responsible for the activation of civic motivation, particularly among younger individuals (Boulianne & Shehata, 2022), a pattern that is also confirmed by our data. Furthermore, because studies examining dutiful citizenship and news consumption rely on diverse and partly non-overlapping item batteries (Choi, 2024; Copeland & Feezell, 2017; Leißner et al., 2019; McCombs & Poindexter, 1983; Ohme, 2019; Ohme et al., 2022), important distinctions may be obscured, such as activating versus suppressing components of civic duty and the moderating role of political interest.

The findings suggest promising avenues for future research on the contemporary role and conception of dutiful citizenship and news consumption, including investigating why people with a weaker sense of duty to stay informed find GenAI more appealing for news consumption than other groups. One possible explanation for this negative association lies in the normative nature of informational duty. A strong sense of duty to stay informed reflects a self-directed approach to news consumption grounded in the civic obligation to actively seek out news about the world (McCombs & Poindexter, 1983). By contrast, GenAI automates information selection, delivery, and potentially even interpretation, emphasizing convenience rather than active engagement. Individuals with a strong sense of informational duty may therefore be less inclined to rely on GenAI for political information.

Regarding the intentional avoidance of news, which had no significant association with using GenAI for political information and news in our study, the results point in a similar direction to those found for low news consumers: people who report high intentional news avoidance do not increasingly turn to GenAI for news. Accordingly, although the affordances of GenAI address some of the obstacles identified as causes of news avoidance, we did not see this potential materialize. However, only one of the three main predictors of intentional news avoidance, namely, media trust, was controlled for. Thus, future studies should also examine users’ perceived information overload and its effects on their subjective well-being when using GenAI for news for a more nuanced understanding of the matter.

Although hopes for GenAI to transform news consumption have largely not materialized, there is also no evidence that fears of GenAI displacing traditional journalistic news are justified. On the contrary, the findings suggest that those who use GenAI heavily for news tend to have a richer overall news diet that includes journalistic sources. In this sense, GenAI appears to complement rather than replace journalistic news consumption, providing news enthusiasts with yet another way to stay informed. However, whether journalistic content appears in GenAI outputs, how it appears, and which content is selected, including the preferential treatment of certain providers (Schatto-Eckrodt et al., 2025), could reshape these conditions in the future and potentially influence user behavior, warranting further research.

This study operationalized news use via GenAI in three ways: as the consumption of general political information, as a substitute for traditional news, and as information related to the 2024 European elections. Contrary to our expectation, this differentiation revealed little nuance. The consistent effects we observed tended to be either highly significant with strong magnitudes or not significant, while the few inconsistent exceptions were comparatively weak and only marginally significant, making their interpretation speculative. Additional research is therefore needed to develop a more robust understanding, depending on the specific focus of interest.

Although this study was conducted during the early stages of GenAI technology’s development, the prevalence of general GenAI use identified in the study demonstrates that technology diffusion has already moved beyond early adopters. However, its use for political information and news has not seen similar uptake, and its future trajectory in this regard remains uncertain. Routines of news consumption and patterns of news socialization will likely transform over time. For instance, as GenAI use continues to diffuse widely, prominent usage motives like fun and amusement (Skjuve et al., 2024) may lead to spillover and familiarization effects that extend to news consumption. Furthermore, some users may experience a sense of self-efficacy through GenAI’s accessible, conversational, and hyper-personalized format, which might enable them to understand and interact with complex political issues and terms in unprecedented ways. Although this technology is widely accessible due to its intuitive usage and often cost-free entry points, meaningful use cases are likely to emerge gradually over time and through experience. These shifts, especially in news consumption, may only become visible over the long term and be driven more by cohort differences than by rapid behavioral change (Peiser, 2000).

Shifts may occur not only over time but also as a result of adjustments to the technology. Consequently, because platform affordances shape news consumption and engagement (Dvir-Gvirsman et al., 2024; Lou et al., 2021; S. Park et al., 2021), both deliberate and inadvertent changes could make GenAI more attractive for such purposes, particularly in a rapidly evolving field. Tapping into the diffusion of innovation theory and going beyond the temporal dimension (Rogers, 2003), a promising yet currently missing aspect may be a greater social integration or functionality of GenAI that enhances observability and social relevance. Currently, GenAI use is a rather solitary experience, driven by active user input and disconnected from a social experience. Relatedly, integrating GenAI more closely into familiar contexts could reduce perceived usage complexity and increase opportunities for trialability. Both aspects could be addressed through greater integration of GenAI into social media, for example, by enabling users to probe GenAI with follow-up questions about viewed posts or by showing what others have asked about the same content. Such integration could bring the affordances of GenAI, most notably tailor-made dialogue-style interaction, directly into existing media practices, particularly among users with lower levels of traditional news consumption. This may enhance the transformational potential of GenAI.

The study has a few limitations. First, the findings rely on self-reported data, which is often imprecise and subject to bias (Parry et al., 2021; Prior, 2009; Scharkow, 2019). To mitigate this, future research should employ tracking and automated content analysis (Reiss, 2023) to assess actual usage patterns of GenAI. This approach would provide a more accurate and comprehensive understanding of how GenAI is utilized in the context of political information and news, including the extent, manner, and content of its use. Second, although the results are representative of the online population in Germany, they may not be generalizable to other countries. Notably, different attitudes toward technology, GenAI, and privacy could influence the use of this technology for political information and news in other cultural contexts. Finally, the observational study design reflects associations rather than causality, leaving questions such as the effects of GenAI use for news on journalistic news consumption up to future research.

9. Conclusion

This study is one of the first to investigate the relationship between GenAI and political information and news consumption from a user perspective and among the first to develop the concept of affordances of GenAI in the context of news and political information. The findings suggest that, at present, GenAI does not play a significant role in political information and news consumption. Moreover, the evidence from this study points against any disruptive potential of GenAI and the affordances it provides in the context of news and political information. In particular, there is currently no indication that using GenAI for news is associated with the displacement of journalistic news consumption.

However, the technology is still in the early stages of adoption, and its development, its acceptance by users, and its integration into everyday life are dynamic. Future research should follow the adoption process by employing various methodological approaches to deepen our understanding of GenAI in the context of news consumption.

Funding

This work was supported by the German Federal Ministry for Education and Research under grant number 16DWCQP07. The funder had no role in the design, conduct, or reporting of the study.

Supplementary Material

Supplemental material and the full data and code to support the findings of this study are openly available at https://osf.io/7gcmy/.

References

Aalberg, T., Blekesaune, A., & Elvestad, E. (2013). Media Choice and Informed Democracy: Toward Increasing News Consumption Gaps in Europe? The International Journal of Press/Politics, 18(3), 281 – 303. https://doi.org/10.1177/1940161213485990

Altay, S., Nielsen, R. K., & Fletcher, R. (2023). News Can Help! The Impact of News Media and Digital Platforms on Awareness of and Belief in Misinformation. The International Journal of Press/Politics, Article 194016122211489. https://doi.org/10.1177/19401612221148981

Araujo, T., Helberger, N., Kruikemeier, S., & De Vreese, C. H. (2020). In AI we trust? Perceptions about automated decision-making by artificial intelligence. AI & SOCIETY, 35(3), 611–623. https://doi.org/10.1007/s00146-019-00931-w

Augenstein, I., Baldwin, T., Cha, M., Chakraborty, T., Ciampaglia, G. L., Corney, D., DiResta, R., Ferrara, E., Hale, S., Halevy, A., Hovy, E., Ji, H., Menczer, F., Miguez, R., Nakov, P., Scheufele, D., Sharma, S., & Zagni, G. (2024). Factuality Challenges in the Era of Large Language Models and Opportunities for Fact-Checking. Nature Machine Intelligence, 6(8), 852–863. https://doi.org/10.1038/s42256-024-00881-z

Ayre, J., Cvejic, E., & McCaffery, K. J. (2025). Use of CHATGPT to obtain health information in Australia, 2024: Insights from a nationally representative survey. Medical Journal of Australia, 222(4), 210–212. https://doi.org/10.5694/mja2.52598

Bautista, J. R., Herbert, D., Farmer, M., De Torres, R. Q., Soriano, G. P., & Ronquillo, C. (2025). Health consumers’ use and perceptions of health information from generative artificial intelligence chatbots: A scoping review. Applied Clinical Informatics. https://doi.org/10.1055/a-2647-1210

Behre, J., Hölig, S., Stöwing, E., & Möller, J. (2025). Reuters Institute Digital News Report 2025: Ergebnisse für Deutschland. Arbeitspapiere des Hans-Bredow-Instituts, 77, 75 S. https://doi.org/10.21241/SSOAR.102887

Betakova, D., Boomgaarden, H., Lecheler, S., & Schäfer, S. (2025). I do not (want to) know! The relationship between intentional news avoidance and low news consumption. Mass Communication and Society, 28(3), 413–440. https://doi.org/10.1080/15205436.2024.2304759

Blassnig, S., Mitova, E., Pfiffner, N., & Reiss, M. V. (2023). Googling referendum campaigns: Analyzing online search patterns regarding Swiss direct-democratic votes. Media and Communication, 11(1), 19–30. https://doi.org/10.17645/mac.v11i1.6030

Boczkowski, P. J., Mitchelstein, E., & Matassi, M. (2018). “News comes across when I’m in a moment of leisure”: Understanding the practices of incidental news consumption on social media. New Media & Society, 20(10), 3523 – 3539. https://doi.org/10.1177/1461444817750396

Boukes, M., & Vliegenthart, R. (2017). News consumption and its unpleasant side effect: Studying the effect of hard and soft news exposure on mental well-being over time. Journal of Media Psychology, 29(3), 137–147. https://doi.org/10.1027/1864-1105/a000224

Boulianne, S., & Shehata, A. (2022). Age differences in online news consumption and online political expression in the United States, United Kingdom, and France. The International Journal of Press/Politics, 27(3), 763–783. https://doi.org/10.1177/19401612211060271

Chatterji, A., Cunningham, T., Deming, D., Hitzig, Z., Ong, C., Shan, C., & Wadman, K. (2025). How people use ChatGPT. OpenAI. https://cdn.openai.com/pdf/a253471f-8260-40c6-a2cc-aa93fe9f142e/economic-research-chatgpt-usage-paper.pdf

Choi, H. (2024). Exploring Public Perception of News and Citizenship and Its Relationship with News Use. Journalism Practice, 1–19. https://doi.org/10.1080/17512786.2024.2407359

Copeland, L., & Feezell, J. T. (2017). The Influence of Citizenship Norms and Media Use on Different Modes of Political Participation in the US. Political Studies, 65(4), 805–823. https://doi.org/10.1177/0032321717720374

Dahlgren, P. M. (2019). Selective Exposure to Public Service News Over Thirty Years: The Role of Ideological Leaning, Party Support, and Political Interest. The International Journal of Press/Politics, 24(3), 293–314. https://doi.org/10.1177/1940161219836223

Damstra, A., Vliegenthart, R., Boomgaarden, H., Glüer, K., Lindgren, E., Strömbäck, J., & Tsfati, Y. (2023). Knowledge and the news: An investigation of the relation between news use, news avoidance, and the presence of (mis)beliefs. The International Journal of Press/Politics, 28(1), 29–48. https://doi.org/10.1177/19401612211031457

Delli Carpini, M. X. (2000). In Search of the Informed Citizen: What Americans Know About Politics and Why It Matters. The Communication Review, 4(1), 129–164. https://doi.org/10.1080/10714420009359466

Dodds, T., Zamith, R., & Lewis, S. C. (2026). The AI turn in journalism: Disruption, adaptation, and democratic futures. Journalism, 27(3), 530 – 544. https://doi.org/10.1177/14648849251343518

Dvir-Gvirsman, S., Sude, D., & Raisman, G. (2024). Unpacking news engagement through the perceived affordances of social media: A cross-platform, cross-country approach. New Media & Society, 26(11), 6487–6509. https://doi.org/10.1177/14614448231154432

Edgerly, S., Vraga, E. K., Bode, L., Thorson, K., & Thorson, E. (2018). New media, new relationship to participation? A closer look at youth news repertoires and political participation. Journalism & Mass Communication Quarterly, 95(1), 192 – 212. https://doi.org/10.1177/1077699017706928

Edison, S. W., & Geissler, G. L. (2003). Measuring attitudes towards general technology: Antecedents, hypotheses and scale development. Journal of Targeting, Measurement and Analysis of Marketing, 12(2), 137–156.

Evans, S. K., Pearce, K. E., Vitak, J., & Treem, J. W. (2017). Explicating affordances: A conceptual framework for understanding affordances in communication research. Journal of Computer-Mediated Communication, 22(1), 35–52. https://doi.org/10.1111/jcc4.12180

Ferrara, E. (2023). Should ChatGPT be biased? Challenges and risks of bias in large language models. First Monday, 28(11). https://doi.org/10.5210/fm.v28i11.13346

Frewer, L. J., Howard, C., & Shepherd, R. (1998). Understanding public attitudes to technology. Journal of Risk Research, 1(3), 221–235. https://doi.org/10.1080/136698798377141

Gao, S., Fang, J., Tu, Q., Yao, Z., Chen, Z., Ren, P., & Ren, Z. (2024). Generative News Recommendation. Proceedings of the ACM Web Conference 2024, 3444–3453. https://doi.org/10.1145/3589334.3645448

Gibson, J. J. (2014). The Ecological Approach to Visual Perception. Psychology Press. https://doi.org/10.4324/9781315740218

Gil de Zúñiga, H., & Diehl, T. (2019). News finds me perception and democracy: Effects on political knowledge, political interest, and voting. New Media & Society, 21(6), 1253–1271. https://doi.org/10.1177/1461444818817548

Gil de Zúñiga, H., Strauss, N., & Huber, B. (2020). The proliferation of the “News finds me” perception across societies. International Journal of Communication, 14, 1605–1633.

Gil de Zúñiga, H., Weeks, B., & Ardèvol-Abreu, A. (2017). Effects of the news-finds-me perception in communication: Social media use implications for news seeking and learning about politics: NEWS FINDS ME PERCEPTION. Journal of Computer-Mediated Communication, 22(3), 105 – 123. https://doi.org/10.1111/jcc4.12185

Gillespie, N., Lockey, S., & Curtis, C. (2021). Trust in Artificial Intelligence: A five country study. The University of Queensland and KPMG. https://doi.org/10.14264/e34bfa3

Goyanes, M., Ardèvol-Abreu, A., & Gil de Zúñiga, H. (2023). Antecedents of news avoidance: Competing effects of political interest, news overload, trust in news media, and “news finds me” perception. Digital Journalism, 11(1), 1–18. https://doi.org/10.1080/21670811.2021.1990097

Greussing, E., Guenther, L., Baram-Tsabari, A., Dabran-Zivan, S., Jonas, E., Klein-Avraham, I., Taddicken, M., Agergaard, T. E., Beets, B., Brossard, D., Chakraborty, A., Fage-Butler, A., Huang, C.-J., Kankaria, S., Lo, Y.-Y., Nielsen, K. H., Riedlinger, M., & Song, H. (2025). The perception and use of generative AI for science-related information search: Insights from a cross-national study. Public Understanding of Science, 34(5), 599–615. https://doi.org/10.1177/09636625241308493

Habermas, J. (1989). The structural transformation of the public sphere: An inquiry into a category of bourgeois society. MIT Press.

Holton, A. E., & Chyi, H. I. (2012). News and the overloaded consumer: factors influencing information overload among news consumers. Cyberpsychology, Behavior, and Social Networking, 15(11), 105 – 123. https://doi.org/10.1089/cyber.2011.0610

Humprecht, E., Esser, F., & Van Aelst, P. (2020). Resilience to online disinformation: A framework for cross-national comparative research. The International Journal of Press/Politics, 25(3), 493–516. https://doi.org/10.1177/1940161219900126

Just, N., & Latzer, M. (2017). Governance by algorithms: Reality construction by algorithmic selection on the Internet. Media, Culture & Society, 39(2), 238–258. https://doi.org/10.1177/0163443716643157

Kacperski, C., Ulloa, R., Bonnay, D., Kulshrestha, J., Selb, P., & Spitz, A. (2024). Who are the users of ChatGPT? Implications for the digital divide from web tracking data (arXiv:2309.02142). arXiv. http://arxiv.org/abs/2309.02142

Kai-ming Au, A., & Enderwick, P. (2000). A cognitive model on attitude towards technology adoption. Journal of Managerial Psychology, 15(4), 266–282. https://doi.org/10.1108/02683940010330957

Karlsen, R., Beyer, A., & Steen-Johnsen, K. (2020). Do high-choice media environments facilitate news avoidance? A longitudinal study 1997-2016. Journal of Broadcasting & Electronic Media, 64(5), 794 – 814. https://doi.org/10.1080/08838151.2020.1835428

Kelly, D., & Fu, X. (2007). Eliciting better information need descriptions from users of information search systems. Information Processing & Management, 43(1), 30–46. https://doi.org/10.1016/j.ipm.2006.03.006

Kümpel, A. S. (2020). The Matthew Effect in social media news use: Assessing inequalities in news exposure and news engagement on social network sites (SNS). Journalism, 21(8), 1083–1098. https://doi.org/10.1177/1464884920915374

Lecheler, S., & de Vreese, C. H. (2017). News media, knowledge, and political interest: Evidence of a dual role from a field experiment: News media, knowledge, and political interest. Journal of Communication, 67(4), 545–564. https://doi.org/10.1111/jcom.12314

Leeper, T. J. (2020). Raising the floor or closing the gap? How media choice and media content impact political knowledge. Political Communication, 37(5), 719 – 740. https://doi.org/10.1080/10584609.2020.1753866

Leißner, L., Valentim, A., Porten-Cheé, P., & Emmer, M. (2019). The selective catalyst: Internet use as a mediator of citizenship norms’ effects on political participation (Weizenbaum Series 1). Weizenbaum Institute https://doi.org/10.34669/WI.WS/1

Lind, F., & Boomgaarden, H. G. (2019). What we do and don’t know: A meta-analysis of the knowledge gap hypothesis. Annals of the International Communication Association, 43(3), 210–224. https://doi.org/10.1080/23808985.2019.1614475

Lipka, M., & Eddy, K. (2025, October 1). Relatively few Americans are getting news from AI chatbots like ChatGPT. Pew Research Center. https://www.pewresearch.org/short-reads/2025/10/01/relatively-few-americans-are-getting-news-from-ai-chatbots-like-chatgpt/

Logg, J. M., Minson, J. A., & Moore, D. A. (2019). Algorithm appreciation: People prefer algorithmic to human judgment. Organizational Behavior and Human Decision Processes, 151, 90–103. https://doi.org/10.1016/j.obhdp.2018.12.005

Lou, C., Tandoc Jr., E. C., Hong, L. X., Pong, X. Y., Lye, W. X., & Sng, N. G. (2021). When motivations meet affordances: News consumption on Telegram. Journalism Studies, 22(7), 934–952. https://doi.org/10.1080/1461670X.2021.1906299

Luskin, R. C. (1990). Explaining political sophistication. Political Behavior, 12(4), 331–361. https://doi.org/10.1007/BF00992793

McCombs, M., & Poindexter, P. (1983). The duty to keep informed: News exposure and civic obligation. Journal of Communication, 33(2), 88–96. https://doi.org/10.1111/j.1460-2466.1983.tb02391.x

Merten, L., Metoui, N., Makhortykh, M., Trilling, D., & Möller, J. (2022). News won’t find me? Exploring inequalities in social media news use with tracking data. International Journal of Communication, 16, 1127–1147.

Mont’Alverne, C., Arguedas, A. R., Banerjee, S., Toff, B., Fletcher, R., & Nielsen, R. K. (2024). The electoral misinformation nexus: How news consumption, platform use, and trust in news influence belief in electoral misinformation. Public Opinion Quarterly, 88(SI), 681–707. https://doi.org/10.1093/poq/nfae019

Monzer, C., Moeller, J., Helberger, N., & Eskens, S. (2020). User perspectives on the news personalization process: Agency, trust and utility as building blocks. Digital Journalism, 8(9), 1142–1162. https://doi.org/10.1080/21670811.2020.1773291

Morosoli, S., Resendez, V., Naudts, L., Helberger, N., & De Vreese, C. (2025). “I resist.” A study of individual attitudes towards generative AI in journalism and acts of resistance, risk perceptions, trust and credibility. Digital Journalism, 1–20. https://doi.org/10.1080/21670811.2024.2435579

Newman, N., Ross Arguedas, A., Robertson, C. T., Nielsen, R. K., & Fletcher, R. (2025). Reuters Institute Digital News Report 2025. Reuters Institute for the Study of Journalism.

Nishal, S., & Diakopoulos, N. (2024). Envisioning the Applications and Implications of Generative AI for News Media (arXiv:2402.18835). arXiv. https://doi.org/10.48550/arXiv.2402.18835

Ohme, J. (2019). Updating citizenship? The effects of digital media use on citizenship understanding and political participation. Information, Communication & Society, 22(13), 1903–1928. https://doi.org/10.1080/1369118X.2018.1469657

Ohme, J., Araujo, T., Zarouali, B., & de Vreese, C. H. (2022). Frequencies, drivers, and solutions to news non-attendance: Investigating differences between low news usage and news (topic) avoidance with conversational agents. Journalism Studies, 23(12), 414 – 435. https://doi.org/10.1080/1461670X.2022.2102533

Overgaard, C. S. B. (2023). Mitigating the consequences of negative news: How constructive journalism enhances self-efficacy and news credibility. Journalism, 24(7), 1424–1441. https://doi.org/10.1177/14648849211062738

Palmer, R., & Edgerly, S. (2024). How journalists perceive news avoidance: Reactions and solutions to the missing audience as boundary work. Journalism Studies, 25(12), 1555–1572. https://doi.org/10.1080/1461670X.2024.2345672

Palmer, R., & Toff, B. (2020). What does it take to sustain a news habit? The role of civic duty norms and a connection to a “news community” among news avoiders in the UK and Spain. International Journal of Communication, 14, 1634–1653.

Palmer, R., Toff, B., & Nielsen, R. K. (2020). “The Media covers up a lot of things”: Watchdog ideals meet folk theories of journalism. Journalism Studies, 21(14), 1973 –1989. https://doi.org/10.1080/1461670X.2020.1808516

Palmer, R., Toff, B., & Nielsen, R. K. (2023). Examining assumptions around how news avoidance gets defined: The importance of overall news consumption, intention, and structural inequalities. Journalism Studies, 24(6), 697–714. https://doi.org/10.1080/1461670X.2023.2183058

Park, C. S., & Kaye, B. K. (2020). What’s this? Incidental exposure to news on social media, news-finds-me perception, news efficacy, and news consumption. Mass Communication and Society, 23(2), 157–180. https://doi.org/10.1080/15205436.2019.1702216

Park, S., Sang, Y., Jung, J., & Stroud, N. J. (2021). News engagement: The Roles of technological affordance, emotion, and social endorsement. Digital Journalism, 9(8), 1007–1017. https://doi.org/10.1080/21670811.2021.1981768

Parry, D. A., Davidson, B. I., Sewall, C. J. R., Fisher, J. T., Mieczkowski, H., & Quintana, D. S. (2021). A systematic review and meta-analysis of discrepancies between logged and self-reported digital media use. Nature Human Behaviour, 5(11), 1535–1547. https://doi.org/10.1038/s41562-021-01117-5

Peiser, W. (2000). Cohort replacement and the downward trend in newspaper readership. Newspaper Research Journal, 21(2), 11–22. https://doi.org/10.1177/073953290002100202

Perez, S. (2025, July 2). ChatGPT referrals to news sites are growing, but not enough to offset search declines. Yahoo Finance. https://finance.yahoo.com/news/chatgpt-referrals-news-sites-growing-174840587.html

Peters, C., Schrøder, K. C., Lehaff, J., & Vulpius, J. (2022). News as they know it: Young adults’ information repertoires in the digital media landscape. Digital Journalism, 10(1), 62 – 86. https://doi.org/10.1080/21670811.2021.1885986

Prior, M. (2005). News vs. entertainment: How increasing media choice widens gaps in political knowledge and turnout. American Journal of Political Science, 49(3), 577 – 592.

Prior, M. (2007). Post-broadcast democracy: How media choice increases inequality in political involvement and polarizes elections. Cambridge University Press.

Prior, M. (2009). Improving media effects research through better measurement of news exposure. The Journal of Politics, 71(3), 893–908. https://doi.org/10.1017/S0022381609090781

Prior, M. (2010). You’ve either got it or you don’t? The stability of political interest over the life cycle. The Journal of Politics, 72(3), 747–766. https://doi.org/10.1017/S0022381610000149

Puschmann, C. (2019). Beyond the bubble: Assessing the diversity of political search results. Digital Journalism, 7(6), 824–843. https://doi.org/10.1080/21670811.2018.1539626

Radlinski, F., & Craswell, N. (2017). A theoretical framework for conversational search. Proceedings of the 2017 Conference on Conference Human Information Interaction and Retrieval, 117–126. https://doi.org/10.1145/3020165.3020183

Reid, L. (2024). Generative AI in Search: Let Google do the searching for you. https://blog.google/products/search/generative-ai-google-search-may-2024/

Reisner, A. (2025, June 25). The end of publishing as we know it. The Atlantic. https://www.theatlantic.com/technology/archive/2025/06/generative-ai-pirated-articles-books/683009/

Reiss, M. V. (2023). Dissecting non-use of online news: Systematic evidence from combining tracking and automated text classification. Digital Journalism, 11(2), 363 – 383. https://doi.org/10.1080/21670811.2022.2105243

Reiss, M. V., Festic, N., Latzer, M., & Rüedy, T. (2021). The relevance internet users assign to algorithmic-selection applications in everyday life. Studies in Communication Sciences, 21(1), 71–90. https://doi.org/10.24434/j.scoms.2021.01.005

Reiss, M. V., Knor, E. L., Stöwing, E., Merten, L., & Möller, J. (2025). Zwischen Neugier und Skepsis: Nutzung und Wahrnehmung generativer KI zur Informationssuche in Deutschland. Arbeitspapiere des Hans-Bredow-Instituts, 76, 39 S. https://doi.org/10.21241/SSOAR.100907

Rogers, E. M. (2003). Diffusion of innovations (5th ed.). Free Press.

Scharkow, M. (2019). The reliability and temporal stability of self-reported media exposure: A meta-analysis. Communication Methods and Measures, 13(3), 363 –383. https://doi.org/10.1080/19312458.2019.1594742

Schatto-Eckrodt, T., Liebig, L., Reiss, M. V., Geislinger, R., Schaetz, N., Merten, L., Schröder, J. T., Königslöw, K. K., Laugwitz, L., Knor, E. L., Semmann, M., Behre, J., & Lischka, J. A. (2025). ChatGPT as a news recommender system: Measuring source types and diversity across different interfaces. SocArXiv. https://doi.org/10.31235/osf.io/wjzp3_v3

Schmitt, J. B., Debbelt, C. A., & Schneider, F. M. (2018). Too much information? Predictors of information overload in the context of online news exposure. Information, Communication & Society, 21(8), 1151–1167. https://doi.org/10.1080/1369118X.2017.1305427

Searles, K., & Feezell, J. T. (2023). Scrollability: A new digital news affordance. Political Communication, 40(5), 670–675. https://doi.org/10.1080/10584609.2023.2208083

Sidoti, O., & McClain, C. (2025, June 25). 34% of U.S. adults have used ChatGPT, about double the share in 2023. Pew Research Center. https://www.pewresearch.org/short-reads/2025/06/25/34-of-us-adults-have-used-chatgpt-about-double-the-share-in-2023/

Simmons, G. (2022). Moral mimicry: Large language models produce moral rationalizations tailored to political identity (Version 2). arXiv. https://doi.org/10.48550/ARXIV.2209.12106

Skjuve, M., Brandtzaeg, P. B., & Følstad, A. (2024). Why do people use ChatGPT? Exploring user motivations for generative conversational AI. First Monday, 29(1). https://doi.org/10.5210/fm.v29i1.13541

Skovsgaard, M., & Andersen, K. (2020). Conceptualizing news avoidance: Towards a shared understanding of different causes and potential solutions. Journalism Studies, 21(4), Article 4. https://doi.org/10.1080/1461670X.2019.1686410

Skovsgaard, M., Shehata, A., & Strömbäck, J. (2016). Opportunity structures for selective exposure: Investigating selective exposure and learning in Swedish election campaigns using panel survey data. The International Journal of Press/Politics, 21(4), 1172 – 1190. https://doi.org/10.1177/1940161216658157

Song, H., Jung, J., & Kim, Y. (2017). Perceived news overload and its cognitive and attitudinal consequences for news usage in South Korea. Journalism & Mass Communication Quarterly, 94(4), 1172 – 1190. https://doi.org/10.1177/1077699016679975

Strömbäck, J., Djerf-Pierre, M., & Shehata, A. (2013). The dynamics of political interest and news media consumption: A longitudinal perspective. International Journal of Public Opinion Research, 25(4), 414 – 435. https://doi.org/10.1093/ijpor/eds018

Strömbäck, J., Tsfati, Y., Boomgaarden, H., Damstra, A., Lindgren, E., Vliegenthart, R., & Lindholm, T. (2020). News media trust and its impact on media use: Toward a framework for future research. Annals of the International Communication Association, 44(2),139 – 156. https://doi.org/10.1080/23808985.2020.1755338

Strömbäck, J., Wikforss, Å., Glüer, K., Lindholm, T., & Oscarsson, H. (2022). Knowledge resistance in high-choice information environments. Routledge. https://doi.org/10.4324/9781003111474

Thorson, K. (2020). Attracting the news: Algorithms, platforms, and reframing incidental exposure. Journalism, 21(8), 1067–1082. https://doi.org/10.1177/1464884920915352

Toff, B., & Kalogeropoulos, A. (2020). All the news that’s fit to ignore. Public Opinion Quarterly, 84(S1), 366 – 390. https://doi.org/10.1093/poq/nfaa016

Toff, B., & Nielsen, R. K. (2022). How news feels: Anticipated anxiety as a factor in news avoidance and a barrier to political engagement. Political Communication, 39(6), 697–714. https://doi.org/10.1080/10584609.2022.2123073

Toff, B., & Simon, F. M. (2025). “Or they could just not use it?”: The dilemma of AI disclosure for audience trust in news. The International Journal of Press/Politics, 30(4), 881 – 903. https://doi.org/10.1177/19401612241308697

Trilling, D., & Schoenbach, K. (2013). Skipping current affairs: The non-users of online and offline news. European Journal of Communication, 28(1), 35 – 51. https://doi.org/10.1177/0267323112453671

Wang, S., & Huang, G. (2024). The impact of machine authorship on news audience perceptions: A meta-analysis of experimental studies. Communication Research, 51(7), 815–842. https://doi.org/10.1177/00936502241229794

Wu, C., Wu, F., Huang, Y., & Xie, X. (2022). Personalized news recommendation: Methods and challenges (arXiv:2106.08934). arXiv. https://doi.org/10.48550/arXiv.2106.08934

Xavier, B. (2025). Biases within AI: Challenging the illusion of neutrality. AI & SOCIETY, 40(3), 1545–1546. https://doi.org/10.1007/s00146-024-01985-1

Date received: 21 October 2025

Date accepted: 20 February 2026

---

1. 1 Before forming the final sample, 187 respondents were excluded for monotone answering behavior, excessively short completion times, or failing an attention check. See in the Online Appendix for details on the exclusion criteria.

2. 2 The survey also included a survey experiment (not part of this study), which was approved by the ethics committee of the Faculty of Business, Economics, and Social Sciences of the University of Hamburg (No. 2024-013). For the remainder of the survey (i.e., what is relevant to this study), no ethics approval was obtained, in accordance with German law and research ethics guidelines by the ethics committee and other bodies such as the WZB Research Ethics Policy and Procedures and the German Data Forum’s Principles and Review Procedures of Research Ethics in the Social and Economic Sciences.

3. 3 The Online Appendix and the reproduction material (full data and code), as well as additional robustness checks on the analyses, can be found online at https://osf.io/7gcmy/.