• 55% of those who reject automated news access information primarily through social networks

• 38% of users who prefer news produced by human journalists pay to access digital information

• 52% of under-35s consider automated news to be cheaper than news generated by human journalists

The irruption of artificial intelligence (AI) in the field of journalism has generated new opportunities, but also complex challenges related to its use and scope. The 2024 report studied the attitude of audiences towards the use of generative AI and the Degree conformity with the subject matter of the news. There was less acceptance of automated news when it dealt with serious and relevant topics, compared to those dealing with entertainment or human interest content.

In the year 2025, compliance levels with AI-generated news have been reanalyzed in two modes: news created largely by artificial intelligence with some human oversight(automated news) and news created largely by a human journalist with some financial aid from an artificial intelligence(human news). As the global report data indicates, more and more journalists are using generative AI in newsrooms to support their work. Uses include documentation, research, transcription, translation, headline suggestions or summary information, among other purposes. It is also becoming increasingly common to incorporate automatically generated stories, albeit under human supervision.

Public attitudes towards the use of artificial intelligence (AI) in journalism.

Automated news, largely produced by artificial intelligence (AI) with some human supervision, provokes either discomfort (47%) or indifference (33%) among respondents. Only 13% declare a favorable attitude towards this subject of content. In contrast, the human element remains a core topic in Spaniards' attitude towards information: 26% are comfortable with content produced by an AI-assisted professional, 38% are indifferent and 30% are uncomfortable. These data highlight the confidence that the figure of the journalist as a guarantor of rigor and context still arouses, even when using artificial intelligence tools. The role of the professional in the production of news content significantly reduces both rejection and neutrality, and roughly doubles the audience's comfort level with a fully automated model .

Spanish audience's discomfort with AI-generated automated news is growing

If we analyze in detail the level of audience compliance with automated news (created by artificial intelligence with some human supervision) we can appreciate differences by gender and age. In general, there has been an increase in discomfort with this practice.

As can be seen in the graph, 51% of women express discomfort with automated news (seven points more than men). On the other hand, 16% of men say they feel comfortable with these news contents, a percentage that remains stable with respect to last year. As for the perception of neutrality, around 30% of men and women say they feel neither comfortable nor uncomfortable with automated news.

In relation to the age of the respondents, it is observed that approximately half of the participants, in almost all age groups, express discomfort with automated news. This discomfort reaches its highest point in the 55 to 64 age group , where the percentage reaches 50%. In the other age groups, the level of discomfort is slightly lower. It is important to note that dissatisfaction has increased among the younger segments with respect to the previous year. If we look at those who do show acceptance, 19% of people between 18 and 24 approve of automated news, a figure that represents a decrease of three percentage points compared to 2024. Something similar occurs in the 25-34 age group , where approval reaches 17%. Despite the younger population's greater familiarity with the use of chatbots, there is widespread dissatisfaction with content created mostly by artificial intelligence, a trend that extends to all age groups.

In contrast to the above, audience perception tends to be more favorable towards (human) news produced by journalists. Levels of conformity are slightly higher among women (26%) than among men (25%). In contrast to the negative assessment of automated news, in this case the level of discomfort decreases.

This indicates that the audience tends to accept the widespread use of artificial intelligence as a complementary tool for journalists and media organizations. This acceptance is especially reflected in the percentage of neutral responses, with 38% of men and women stating that they are neither comfortable nor uncomfortable with this practice, remaining stable compared to the previous year. This data suggests a growing familiarity and normalization of the role of AI in journalistic processes.

When comparing the perception according to educational level, it is observed that the discomfort with automated news increases as the level of programs of study increases. For example, 42% of the audience with a low educational level expresses discomfort with this subject of news, while among those with a university degree or postgraduate program the percentage rises to 54%. On the other hand, the levels of neutrality are more defined in the groups with leave and average education, with figures exceeding 30%. As for conformity with this news modality , it remains constant at around 13% at all educational levels.

Levels of comfort with human-made news content increase as the level of programs of study increases. Although the differences are not very significant, it is observed that 23% of people with low educational level feel comfortable with this subject of news, compared to 28% of those who have university programs of study or postgraduate program. In general terms, the discomfort generated by human news is considerably lower than that associated with automated content, at around 30%. The most striking feature is observed in the levels of neutrality (those who say they feel neither comfortable nor uncomfortable with human news using AI), since the percentage decreases as the level of programs of study increases. Thus, 32% of those with a university Degree or postgraduate program say they are neutral towards human news, compared to 45% of those with a low educational level. These factors could be related to other variables, such as distrust of the news or the tendency to intentionally avoid certain news content.

Compliance with news content that integrates AI and confidence in the news consumed.

The presence of artificial intelligence in journalistic routines not only affects the way in which news is produced and distributed, but may also condition the trust of the public that receives it. When analyzing the differences between those who declare high trust in the news they consume, relevant contrasts are identified. Thirty-six percent are comfortable with news produced by (human) journalists, while only 19% sample comfortable with fully automated news.

As for the level of discomfort, the percentages remain at 27% for both news generated entirely by AI and those produced by journalists. The main difference is in the Degree of neutrality: 48% of those who trust news the most feel neither comfortable nor uncomfortable with automated news. In contrast, only 32% take that stance for news produced by human journalists.

Attitudes towards information content with AI by subject of access and payment method on the internet

After analyzing the relationship between trust and news consumption, it is essential to explore how the audience accesses information on the Internet. The report distinguishes six main categories: access through news apps or websites; use of search engines (such as Google or Bing); platforms that aggregate news links, such as Google News, Flipboard or News Republic; social networksFacebook, TikTok, Instagram, X, YouTube, etc.); receiving news via newsletters or email alerts; and notifications via mobile devices or tablets (either by SMS, apps, lock screen or notification center).

Focusing on attitudes towards automated news, 55% of those who say they are indifferent to AI-generated content (e.g., via chatbots) access news via social networks. This is followed by users who receive news via email newsletters or alerts (51%), followed by those who use news apps or websites (49%) and those who use search engines (45%). It is important to note that neutrality levels are similar among those who access news through apps or websites for reading news content and those who receive alerts through devices (more than 40% in both cases).

In contrast to the data presented above, in the case of news produced by journalists, the levels of non-conformity remain at around 30% in all modes of access. As noted in relation to automated news, neutrality is also associated with consumption through social networks, although in a considerably lower proportion: 33% compared to 55% recorded for automated news. In addition, 35% of those who are comfortable with news produced by (human) journalists access them via websites or specific news-reading applications.

In addition to the Degree conformity with the two modalities, respondents were also asked about perceptions of some aspects related to trust in AI-generated news content. Specifically, respondents were asked whether they believe that news generated mainly by artificial intelligence with human supervision is more, less or equally accurate, unbiased, transparent and trustworthy than news produced exclusively by journalists.

Although around 40% of respondents sample neutral in their consideration, if we look at those who respond positively (AI generates more accurate, unbiased, transparent and reliable content), differences are observed according to age. In the four aspects evaluated, young people and young adults (especially those under 35 years of age) tend to rate this content more positively than older respondents (especially those over 65).

When considering other aspects such as cost, timeliness and comprehension of automated news according to age, significant differences are observed. For example, 52% of respondents under 35 years of age perceive these news items to be cheaper, while this perception is lower among the adult population, where only 36% share this opinion. As for the perception of topicality and the ease of understanding the news, there is consensus among respondents of different age ranges. On the one hand, nearly 50% of those over 65 consider automated news to be no more or less current or easy to understand than news created by human journalists. On the other hand, 42% of young people under the age of 35 hold the same opinion in both cases.

Personalization with AI: formatting yes, content not so much

The media industry is considering using AI to better tailor news content to the needs of the individual. This year's survey asked about interest in its use to fulfill different personalization functions, and the responses reveal that Spaniards are not very enthusiastic: a significant proportion of respondents are not interested in the use of AI in any of the proposed cases (25%) or do not know if they are relevant (15%). Among the citizens who do show interest, the use of AI in questions related to the format of the information (48% of respondents) and not so much to the content (27%) is valued above all. Those who are most in favor of this personalization in format are those under 55 years of age, especially Internet users between 25 and 34 years of age (57%) and between 45 and 54 years of age (52%). Among those over 55, the percentage drops to 43%. Also particularly interested in using AI to adapt the formal part of the news are Spaniards with high incomes - 55% of them compared to 45% of the group with low incomes - and with higher education - 55% compared to 42% of the group with less education.

The function most valued by respondents - 26% indicate their interest in using AI for this purpose - is the production of summarized versions of articles, which are quicker to read. One out of five (20%) expressed interest in the automatic translation of news from other languages, which would facilitate access to international content. Below 20% are options such as receiving personalized news recommendations or alerts (17%), the creation of a home page tailored to their interests (16%), the use of AI to adapt the language of the news to different reading levels (14%), for example using more or less advanced vocabulary, the use of chatbots to answer questions about the news (13%) or the conversion of articles from text to audio or vice versa (13%) and from text to video or vice versa (11%).

Interest in personalization with AI shows a general trend of de-escalation as the age of citizens increases. Respondents under 35 are more interested in almost all the options proposed than those above that age threshold, but there are certain functions in which the difference is B: news translated from a foreign language (24% vs. 18%), conversion of articles from text to audio or audio to text (21% vs. 11%) and use of an AI chatbot that answers questions about the news (20% vs. 10%). On the other hand, half of the over-65s sample not interested in AI personalization, something perhaps related to lower media and digital literacy: three out of ten (31%) are not interested in any personalization subject and one in five (19%) say they "don't know" if they are interested. In this age group , those who do show interest in using AI to tailor the information they receive prefer options such as summarized versions of articles or news recommendations and alerts based on their interests; they are, in fact, the age group most interested in the latter possibility (21%). The feature that least appeals to them is the use of a chatbot to answer questions about the news (7%, a far cry from the 23% of adults aged 25-34).

Women are less interested in AI personalization - 27% do not sample interest in any of the proposed options - than men (23%). In general, gender is not a determining variable in relation to the preferred subject of adaptation, but there are two options where there is a significant difference between men and women: articles that convert from text to video or vice versa - chosen by 13% of men compared to 9% of women - and the use of a chatbot that answers questions about the news - 14% of men compared to 11% of women.

We have already noted that, in general, the higher the level of income and education, the greater the interest in personalization with AI. And when we go down to the details, we find that the options with the highest dispersion in both cases are accessing summarized versions of news -chosen by 32% of the group with the highest income compared to 21% with a leave income, and by 32% of citizens with higher education compared to 20% with basic training-, the translation of articles from a foreign language - 25% versus 15% in the case of income, and 26% versus 16% in the case of education - and, to a lesser extent, the conversion of news from text to audio or vice versa - 17% versus 12% in relation to income, 18% versus 9% in relation to level of education. In addition, low income and basic education are directly associated with a greater lack of awareness of the possibility of personalizing news with AI: 17% of people with lower income mark the option "don't know", compared to 9% of the group with the highest income, almost twice as many; and in the case of education, 18% compared to 11%.

The high rate of lack of knowledge among the older, lower income and less educated sectors, compared to other groups, deepens the gap detected in the question on media literacy and reminds us once again of the need to promote public policies and educational strategies that promote media and digital literacy, especially among the most vulnerable sectors.
and educational strategies that promote media and digital literacy, especially among the most vulnerable sectors. The ability to understand and use tools such as artificial intelligence depends not only on access to technology, but also on the knowledge needed to take advantage of it. This gap warns of a risk of information exclusion, in which certain social groups could be left out of the new ways of accessing and consuming information, deepening existing inequalities.