Journalistic AI errors and "telephone games": Are we actually still reading real news?

Secret use of AI? The media's great double standards and their true structural flaw

Artificial intelligence hallucinates, invents facts, and threatens the truth – this is the alarming message from many media outlets. But behind this vocal criticism lies a blatant double standard: While newsrooms publicly warn against the flawed technology, recent studies show that 70 percent of journalists are already secretly using these very AI tools in their daily work. The outrage over machine errors distracts from a much deeper, homegrown problem: the decades-old "telephone game" of journalism. Driven by clickbait and the attention economy, news is adopted without verification, contexts are distorted, and facts are twisted. The real danger to public trust is not the introduction of AI itself – it is the clash of unreliable algorithms with a media system whose quality control has long since structurally eroded. This is an in-depth analysis of perverse incentives, dwindling media trust, and the question of why the industry urgently needs to practice genuine source hygiene.

The defective information system: How structural perverse incentives, the game of telephone, and the silent AI invasion are undermining the foundations of public perception

While newsrooms condemn AI hallucinations, they secretly deploy the same technology on a massive scale – overlooking the fact that their own craft has suffered for decades from a structurally ingrained culture of inaccuracy

The public discourse surrounding artificial intelligence in journalism exhibits a peculiar asymmetry. On the one hand, newsrooms, media critics, and journalists' associations loudly warn of AI hallucinations—the phenomenon in which language models produce statistically plausible but factually incorrect content. The word "hallucination" has become the central buzzword of current media discourse. On the other hand, the reality in newsrooms paints a fundamentally different picture: According to the Media Trend Monitor 2025, 70 percent of German journalists already use AI tools in their daily work—for transcriptions, research, summarizing texts, brainstorming, and optimizing articles.

This contradiction is not only remarkable, it's revealing. The same industry that brands AI hallucinations as a fundamental threat to information quality has long since integrated this technology into its own workflow – often without making the extent of this integration transparent to its readers. When AI structures research, pre-writes texts, or analyzes datasets in the background, the public is usually unaware. The outrage over machine errors thus turns out to be selective: what is perceived as an external threat is accepted internally as a useful tool.

Even more revealing is a recent study by the European Broadcasting Union (EBU), which systematically tested the reliability of popular AI systems. The result: ChatGPT, Gemini, and other chatbots fabricate up to 40 percent of their answers and present them as facts. Every second answer from popular chatbots contains significant errors—whether due to outdated sources, imprecise prompts, or so-called hallucinations. These are real and alarming figures. But these figures raise an uncomfortable follow-up question: If the AI ​​that journalists use daily hallucinates in up to 40 percent of its output, what is the actual error rate in the final products created on this basis?

The forgotten structural flaw: The principle of telephone in journalism

Behind the noise of the AI ​​debate lies an older, deeper, and still largely unaddressed problem: the systematic dissemination and distortion of information by the journalistic establishment itself—long before algorithms came into play. This phenomenon is discussed in media studies under various terms, but ultimately describes one and the same mechanism: news is not generated from primary sources, but rather derived from other news. Every intermediate step reduces precision.

The first key mechanism is circular reporting, known in Anglo-Saxon media studies as "false confirmation." It arises when source B adopts information from source A, source C copies this information from B, and finally, source A cites source C as independent confirmation of its own original claim. The superficial impression that several independent sources confirm one and the same thing is deceptive: all trace back to the same, often erroneous, origin. The result is an epistemic illusion—the condensation of a single, potentially flawed statement into a seeming societal consensus.

The second mechanism is closely related: "churnalism," a portmanteau of the English "churn out" (mass production) and "journalism." It describes a form of journalism in which press releases, agency reports, or articles from competing media outlets are rewritten or simply adopted en masse and largely without verification. Under the pressure of the attention economy, click rates, and real-time reporting, churnalism is no longer the exception but has become the norm for large parts of online journalism. In this practice, journalistic word games are played with remarkable speed: An agency report contains an error, and a hundred newsrooms adopt it within minutes without questioning it.

The third mechanism is the secondary source error. This refers to the journalistic practice of not referring to the original source, the primary source, but rather to what another media outlet has already reported about that source. With each intermediate step, the risk increases that nuances will be lost, figures will be taken out of context, or formulations will imperceptibly shift the original message. A study might show a limited correlation under certain conditions; after three rounds of reporting, the headline will present a universally valid causal relationship. The damage rarely lies in the outright lie, but rather in the gradual deviation from the original message.

What the data really says: Perception and reality as two sides of a crisis

Research on media errors and media trust consistently distinguishes methodologically between two phenomena: the actually measurable journalistic error rate, which can be determined in controlled fact-checking studies, and the perceived inaccuracy, which reflects the subjective distrust of the public. Both dimensions are essential for a sound analysis, as both have real consequences. The perceived error rate determines the extent of social damage caused by false reporting—even if the actual error rate is lower. Conversely, a high actual error rate may have little measurable social impact if the public does not recognize it.

A general, scientifically validated error rate for all news content does not exist. However, the available data from audience perception, journalism research, and media trust studies paint a nuanced and sometimes alarming picture that extends across various countries, media formats, and subject areas.

The American measurements: Up to 44 percent perceived inaccuracy

The most detailed quantitative data comes from the USA. A 2018 Gallup/Knight Foundation study provides the most revealing findings. According to the study, US adults estimate that 44 percent of the content in newspapers, television, and radio is inaccurate. The assessments for social media are even more drastic: 64 percent of the content on social platforms is classified as inaccurate by the same respondents, and 65 percent is even considered disinformation – that is, false or misleading information presented as true.

The distribution by political orientation reveals a remarkable pattern. Republicans perceive significantly more bias, inaccuracy, and misinformation in traditional media than Democrats. However, both groups largely agree on social media: Members of both parties rate the amount of problematic content on these platforms as high. This suggests that the loss of trust in social media is a broader, less partisan phenomenon than the loss of trust in traditional media.

At the institutional level, the erosion is dramatic: The vast majority of US adults—including more than nine out of ten Republicans—report having personally lost trust in the news media in recent years. At the same time, 69 percent of those who have lost trust say that this trust could, in principle, be restored—if the media demonstrated accuracy, transparency, and a move away from bias.

The global perspective: When almost every second person notices errors weekly

Globally, the findings paint a consistent picture of structural credibility problems. According to the Reuters Institute Digital News Report 2018, 59 percent of respondents worldwide said their biggest media concern was that facts were being distorted to advance an agenda—a deliberate, targeted error, not mere carelessness. The same study found that 42 percent of respondents had encountered poor journalism in the previous week—inaccurate reporting or misleading headlines. That's almost half of all news consumers who experience specific quality deficiencies on a weekly basis.

The Reuters Institute Digital News Report 2025, which surveyed nearly 100,000 people in 48 countries, shows that this trend is not a passing fad. Globally, more than half of all respondents – 58 percent – ​​said they were concerned about their ability to distinguish between true and false when consuming news online. This figure was highest in the US and Africa at 73 percent; in Western Europe, at 46 percent, it was comparatively lower, but by no means reassuring. According to the same report, the global proportion of people who trust most news most of the time is only 40 percent – ​​a finding that is hardly surprising after years of continuous erosion, but whose implications can hardly be overstated.

Germany between stabilization and structural mistrust

In Germany, current studies paint a more nuanced, but nonetheless deeply troubling picture. The Mainz Longitudinal Study on Media Trust 2024, conducted by Johannes Gutenberg University Mainz, which has been surveying German public opinion on media attitudes annually since 2015, reveals that 47 percent of the population trusts the media on truly important issues such as environmental problems, health risks, or political scandals. A further 34 percent respond with "partly, partly." Conversely, this means that 20 percent of the German population harbors active distrust of the media, while overall trust is far from being shared by a societal majority.

The thematic differentiation is particularly insightful from an analytical perspective. In terms of trust in individual media categories, public broadcasting leads in 2024 with 61 percent – ​​however, this is also the lowest value recorded to date in the long-term comparison. Only three percent of the German population considers social media to be somewhat or completely trustworthy; video platforms like YouTube reach eight percent, and alternative news sites four percent – ​​also the lowest value recorded to date. Public trust is thus concentrated on a small core of established media outlets, while the growing news channels, which are used particularly by the younger generation, enjoy almost no trust.

The WDR study on media credibility in 2025, conducted by Infratest dimap based on a representative survey of 1,319 eligible voters, shows a slight recovery: 61 percent consider the information in German media to be credible – an increase of five percentage points compared to 2023. This upward trend is real, but must be placed in its historical context: The figure is still below the peak reached during the coronavirus pandemic, when trust temporarily rose due to the acute need for information during the crisis and has since eroded. Furthermore, the study reveals significant political divisions: While 92 percent of Green Party supporters trust public broadcasting, only ten percent of AfD supporters do.

The quasi-in-house solution: How Xpert.Digital closes operational gaps in B2B marketing and sales – Smart Content-Driven Business - Image: Xpert.Digital

Xpert.Digital is a data-driven B2B industry hub led by Konrad Wolfenstein . The company acts as an external, quasi-in-house solution for industrial partners, closing operational gaps in marketing, content, and sales – without requiring additional resources on the client side.

More information here:

• The quasi-in-house solution: How Xpert.Digital closes operational gaps in B2B marketing and sales – Smart Content-Driven Business

The problem of motivation: Why the public names the wrong culprits – and why it is still right

Context, clicks, AI: Why journalism is framed incorrectly today

Crucial for structural analysis is not only the how, but also the why of media errors. The project "Trust in Journalism in Media Structural Change," funded by the German Research Foundation (DFG), provides insightful findings in this regard. 72 percent of respondents agreed with the statement that media outlets primarily strive for circulation and ratings – and see this as the main cause of quality deficiencies. Only 24 percent attribute errors primarily to a lack of journalistic competence.

At first glance, one might dismiss this finding as a misperception by the public: journalists are generally trained professionals, and the idea that they report falsely primarily for economic reasons sounds like a conspiracy theory. In reality, however, this public perception contains a kernel of truth. Structural perverse incentives are well-documented in the media industry: headlines that promise more than the article delivers, selective fact-finding to enhance emotional impact, the reduction of complex issues to a clear good-versus-evil dichotomy – all these are errors that do not stem from incompetence, but from the commercial logic of the attention economy. The public may be identifying the wrong level of blame, but it is identifying the correct systemic problem.

In Germany, 42 percent of adult internet users also lack confidence in their ability to distinguish between true and false information – a figure that has increased by five percentage points compared to 2023. This is not a trivial number: it describes a society in which almost half of active online news consumers no longer reliably master the fundamental skill of information processing – the ability to differentiate between fact and error.

Four types of journalistic inaccuracy: When the detail breaks the overall message

Research distinguishes four qualitatively different types of errors, whose impact on public perception and the overall message of a report varies greatly.

Factual errors are the most visible and yet least consequential category: incorrect numbers, dates, names, or locations. They are easily verifiable, rarely intentional, and generally correctable without affecting the article's core message. Contextual errors are more subtle and impactful: correct facts are presented without the necessary context that unlocks their meaning. A percentage without a point of comparison, a study without mentioning its sample size, a quote without the preceding sentence—these are contextual errors that, while not technically false, can fundamentally shift the overall message.

Emphasis errors—misleading headlines, selective opening sentences, and sensationalist framing—are the most common form of journalistic inaccuracy. According to their own accounts, 42 percent of global news consumers encounter them weekly. They don't work through lies, but rather through the control over which aspect of a story is presented as the most important. Finally, there are agenda-distorted errors: the selective choice or distortion of facts to promote a particular viewpoint. This type is the most prevalent global media concern—59 percent of news consumers worldwide cite it as their biggest worry.

Contextual errors and accentuation errors are particularly difficult to quantify because they are rarely recognized as classic fake news. Their impact stems not from a single lie, but from the accumulation of small omissions, emphases, and framing that create a specific image of reality without being factually incorrect at any point. This makes them the most dangerous and, at the same time, the most difficult to prove form of journalistic inaccuracy.

The social media problem: When mistrust migrates into a parallel world

On social media platforms, which have become the primary news source for a growing and demographically young segment of the population, all the problems of traditional journalism—churnalism, circular reporting, contextual errors—are amplified and further exacerbated by algorithmic amplification and the complete elimination of editorial quality control. In Germany, only five percent of the population considers social media platforms credible. TikTok and similar services have a trust rating below ten percent.

Nevertheless, social networks remain the most important news source for 18- to 24-year-olds: a third of this age group cite social media as their primary source of information, and 17 percent receive their news exclusively from there. This creates a structurally explosive situation: a steadily growing segment of the population obtains its daily news from a channel that it itself considers largely unreliable. Trust and usage are far apart. This is not individual irrationality, but rather a consequence of the lack of attractive, trustworthy alternatives in the preferred formats and on the preferred platforms of these target groups.

Added to this is the psychological effect of creating uncertainty: A study on political deepfake videos showed that such content doesn't necessarily deceive users, but leads to greater uncertainty. This uncertainty spills over into general trust in news: Those who regularly encounter manipulated or misleading content on a platform tend to view legitimate information sources there with skepticism as well. The credibility crisis of journalism is not only exacerbated by social media – it is exported to channels where reputable journalism is already at a structural disadvantage.

The new AI paradox: Machine errors and human game of telephone in competition

The widespread use of AI in newsrooms is creating a new, previously little-discussed problem: the overlapping of human and machine error sources. If a journalist uses ChatGPT for research preparation and the system produces up to 40 percent erroneous content, and if this journalist then – as around one-fifth of media professionals admit – doesn't fully verify the output due to lack of time, a new form of journalistic telephone emerges: The AI ​​hallucinates, the human takes over, and the reader believes it.

The irony is complete: Classic telephone journalism works because human editors, under time pressure, adopt content from other sources without verifying it. The AI-powered version operates on the same basic principle—except that the first "source" is now a machine whose relationship to truth is statistical, not epistemic. AI systems don't know what is true. They produce formulations that sound statistically plausible based on their training data. A system that sounds convincing, even if it's hallucinating, is particularly dangerous for uncritical use—because the critical corrective, the skepticism toward the content, is suppressed by the fluent phrasing.

The resulting insight is uncomfortable for the industry: Anti-AI rhetoric in journalism is often less a fundamental rejection of machine errors than a defense against external competition and an identity narrative. The core structural problem—lack of source hygiene, economically motivated abridgment, circular reporting—existed long before AI and is merely scaled up by its use under unfavorable conditions.

A systemic design problem of the attention economy

The available data does not allow for a simple, straightforward answer to the question of a general error rate in journalism. However, it does allow for a structurally clear conclusion: The perceived error and inaccuracy rate ranges from approximately 25 to over 60 percent, depending on the medium, country, and subject area. Crucially, this requires distinguishing between obvious falsehoods and the more subtle, yet more impactful type of contextual error – an error that fundamentally alters the overall message not through lies, but through omission, framing, or a one-sided focus.

This type of error is the most widespread, the most difficult to prove, and the one that most deeply undermines the foundations of the public information space. The fact that 72 percent of the German population cites circulation and ratings pressure as the main cause of quality deficiencies reveals a crucial collective insight: The problem is not the random failure of individual journalists, but a systemic design flaw in the attention-driven media business model. Those who publish under constant click pressure optimize for reach, not for truth. Those who operate under time pressure resort to secondary sources instead of verifying primary sources. Those in competition adopt what their rivals have already published—thus reinforcing precisely the game of telephone that erodes the information quality of the entire system.

The Reuters Institute Digital News Report 2025 shows that trust in news in Germany remains largely stable at 45 percent, but is still below the peak seen during the coronavirus pandemic. Stability at a low level is no cause for complacency. It is a symptom of a structurally damaged relationship between media and the public – a relationship that cannot be repaired by condemning AI hallucinations, but only by what has been neglected for decades: consistent source hygiene, transparency about production processes, and the honest admission that the game of telephone in journalism is not a new invention of the machine.

☑️ Our business language is English or German

☑️ NEW: Correspondence in your native language!

Konrad Wolfenstein

I and my team are happy to be available to you as your personal advisor.

You can contact me by filling out the contact form here [email protected]:or simply call me at +49 7348 4088 965. My email address is

I'm looking forward to our joint project.

☑️ SME support in strategy, consulting, planning and implementation

☑️ Creation or realignment of the digital strategy and digitization

☑️ Expansion and optimization of international sales processes

☑️ Global & Digital B2B trading platforms

☑️ Pioneer Business Development / Marketing / PR / Trade Fairs

B2B support and SaaS for SEO and GEO (AI search) combined: The all-in-one solution for B2B companies - Image: Xpert.Digital

AI search changes everything: How this SaaS solution will revolutionize your B2B ranking forever.

The digital landscape for B2B companies is undergoing rapid change. Driven by artificial intelligence, the rules of online visibility are being rewritten. For companies, it has always been a challenge not only to be visible in the digital mass, but also to be relevant to the right decision-makers. Traditional SEO strategies and managing local presence (geo-marketing) are complex, time-consuming, and often a battle against constantly changing algorithms and intense competition.

But what if there were a solution that not only simplified this process but also made it smarter, more predictive, and far more effective? This is where the combination of specialized B2B support with a powerful SaaS (Software as a Service) platform comes into play, specifically designed for the demands of SEO and GEO in the age of AI search.

This new generation of tools no longer relies solely on manual keyword analysis and backlink strategies. Instead, it leverages artificial intelligence to more accurately understand search intent, automatically optimize local ranking factors, and conduct real-time competitive analysis. The result is a proactive, data-driven strategy that gives B2B companies a decisive advantage: they are not only found, but perceived as the leading authority in their niche and location.

Here's the symbiosis of B2B support and AI-powered SaaS technology that transforms SEO and GEO marketing, and how your company can benefit from it to grow sustainably in the digital space.

More information here:

​

• B2B Support & Blog for SEO, GEO and AIS – Artificial Intelligence Search
• Forget expensive SEO tools – this alternative dominates with unbeatable B2B features