LOOKING BACK | Revving Up a Misinformation Machine: Media, AI and Bias

Across industries and personal use cases, our embrace of AI usually comes down to a central question: Can it be trusted? 

We don't know about you, dear reader, but in our minds, we don't real benchmark AI's trustworthiness against people, we benchmark it against our own abstract notion of the truth. We don't often ask whether we can be trusted to have well-reasoned opinions about AI.

As members of the media, we feel like we don't ask that question enough of ourselves and question whether we can fairly discuss AI, so we're dedicating a Looking Back column to examining the impact of media bias on AI, and the impact of AI bias on media.  

The narratives we build around disruptive technologies directly impact how those technologies will be accepted and adopted, but in AI's case, the ability of generative technology to craft narratives of its own and contribute to its own discourse will shape how, when and why people choose to engage with it. Thus, if we're going to talk about how human biases, like media bias, may shape our use and adoption of AI, we should talk about how AI biases are going to shape those human biases, too. 

And we're going to start with our own industry, because bias should be on the mind of anyone who works in and engages with journalism. Bias leads to sloppy journalism. While a tired journalist might accidentally flip the results of a survey around so that instead of portraying rising AI adoption among financial professionals they appear to show declining AI use, an editor only accepts those results if they affirm an already held, but mistaken belief, that artificial intelligence is bad. 

And who could blame an editor for feeling that way? Any veteran of newsrooms sees all the empty chairs around, and with AI being the rising technology making writers and editors more efficient, it's easier to blame the new big thing than to acknowledge that AI is just finishing off what decades of decline have already done to the news industry. AI didn't kill print, and it didn't kill broadcast journalism, either, but that's not what it feels like when all your friends are losing their careers. In some ways, AI is going to take a lot of the blame for what economics and other technologies have already done.

As artificial intelligence transforms journalism, the news industry faces a complicated crisis of credibility. Journalists are accused of being both too hostile and too friendly toward AI, while AI-generated news raises questions about accuracy, political bias and public trust. For financial services professionals, the debate offers important lessons about the future of information, financial advice and the relationship between human judgment and machine intelligence. 

Is the news media biased against artificial intelligence? The answer depends on which media organizations are being examined, which aspects of artificial intelligence are being covered and, increasingly, whose economic interests are at stake. A technology journalist celebrating a breakthrough in machine reasoning, a labor reporter investigating AI-related layoffs and a financial columnist questioning the valuations of AI companies might all be describing the same technological revolution from very different perspectives. None is necessarily wrong, and none is necessarily unbiased. 

Artificial intelligence presents an unusual challenge for journalism because it is simultaneously a subject of reporting, a potential competitor, an increasingly important tool of the trade and a technology that could fundamentally alter the economics of news production. Media organizations are expected to evaluate AI impartially while facing the possibility that the technology will replace some of their employees, undermine their business models and redirect audiences away from their websites. 

Meanwhile, AI developers are building systems that summarize, interpret and sometimes replace the very journalism that helps train and inform them. Those systems have their own biases, inherited from training data, reinforced through model development and potentially amplified by their interactions with users. 

The resulting relationship is reciprocal: Journalism influences how artificial intelligence understands the world, while artificial intelligence increasingly influences how journalism is produced, distributed and consumed. 

For financial services professionals, this is more than a debate about the future of newspapers. Investment decisions, market sentiment, financial advice and consumer behavior all depend on the quality and credibility of information. If both human and artificial intelligence are susceptible to bias, understanding their respective weaknesses becomes an essential part of financial literacy. 

Is the Media Anti-AI? 

Evidence suggests that media coverage of artificial intelligence varies significantly according to political orientation, editorial priorities and the economic interests of publishers. 

Researchers at Virginia Tech examined the relationship between political ideology and media sentiment toward AI, finding that liberal-leaning news organizations generally expressed more negative sentiment toward the technology than conservative-leaning outlets. The researchers associated some of those differences with competing concerns about inequality, discrimination, innovation and economic development. 

The findings challenge the assumption that the media has a single, consistent attitude toward artificial intelligence. Coverage can reflect broader political and cultural divisions, with some publications emphasizing potential economic benefits and others devoting more attention to social consequences. 

Yet negative coverage does not necessarily demonstrate unfair treatment. Reporting about algorithmic discrimination, intellectual property disputes, environmental costs, misinformation and technological unemployment addresses genuine issues. Journalists have an obligation to investigate those issues even when their reporting damages the reputations of AI developers. 

The distinction between healthy skepticism and anti-AI bias lies primarily in evidence, proportionality and consistency. 

A skeptical journalist questions whether an AI company can deliver on its promises, examines independent performance evaluations, investigates unintended consequences and corrects errors when new information becomes available. An anti-AI journalist might instead assume that the technology is inherently harmful, dismiss legitimate achievements or interpret every failure as proof that the entire field is fundamentally misguided. 

The opposite problem is equally important. Pro-AI bias can manifest as uncritical repetition of corporate announcements, exaggerated predictions about productivity and employment, or the assumption that technological advancement necessarily produces social progress. 

Financial journalism provides a particularly relevant comparison. Reporters covering emerging technologies must distinguish genuine innovation from promotional narratives designed to attract investors. During speculative investment booms, enthusiastic coverage can reinforce market optimism, while excessive skepticism can obscure legitimate commercial opportunities. 

Neither excitement nor apprehension is inherently evidence of bias. The problem arises when either becomes a substitute for independent investigation. 

Can AI Get a Fair Shake in Journalism? 

Journalism's complicated economic relationship with AI makes impartiality especially difficult. 

Traditional news organizations have endured decades of declining print advertising, competition from digital platforms, shrinking local newsrooms and changing consumer habits. Generative AI introduces another potential disruption by allowing technology companies and consumers to produce summaries, articles and other informational products at extremely low marginal cost. 

Publishers are understandably concerned that AI-generated answers will reduce traffic to original reporting, weakening advertising and subscription revenue. Journalists also face uncertainty about whether employers will replace human writers, editors, researchers and production personnel with automated systems. 

These concerns create possible conflicts of interest. A journalist investigating AI-related job displacement may personally face that same risk. A publisher criticizing AI companies for using copyrighted news material may simultaneously be negotiating licensing arrangements with those companies. 

However, economic exposure does not automatically invalidate reporting. Financial journalists routinely cover banks, asset managers and public companies whose activities influence their own employers. The appropriate response is disclosure, editorial independence and rigorous sourcing rather than an assumption that impartiality is impossible. 

AI companies also deserve scrutiny when they accuse journalists of bias. Developers have strong commercial incentives to characterize criticism as exaggerated, technologically uninformed or motivated by hostility to innovation. Such accusations can be legitimate, but they can also function as a way to avoid answering difficult questions. 

The fairest journalism evaluates AI systems according to demonstrated capabilities, limitations and consequences rather than ideological allegiance. It recognizes that a technology can be extraordinarily useful and potentially harmful at the same time. 

Can AI Be Trusted as a Journalism Tool? 

Despite the tensions between journalism and artificial intelligence, AI is already useful in many newsroom operations. 

Automated transcription, translation, document analysis, data visualization and research assistance can reduce the time required to produce complicated stories. Machine learning can identify anomalies in financial disclosures, analyze enormous public databases and help investigative reporters uncover relationships that might otherwise escape attention. 

For financial journalists, AI can examine thousands of regulatory filings, earnings transcripts and economic releases, identifying patterns that warrant further investigation. A human reporter might spend days reviewing documents that an AI system can process in minutes. 

But efficiency is not the same as reliability. 

Generative AI systems can invent quotations, misidentify sources, misunderstand statistical relationships and present unsupported conclusions with extraordinary confidence. Even systems that retrieve information from external sources can misinterpret what those sources actually say. 

The danger is particularly acute when AI-generated content appears polished and authoritative. An inaccurate article written by a chatbot may contain fewer obvious warning signs than an equally inaccurate article produced by an inexperienced human writer. 

There is also the problem of algorithmic fairness. AI models can reflect demographic stereotypes, political assumptions and cultural prejudices present in their training materials. 

Research from the University of Chicago has documented discriminatory associations involving speakers of African American English, while other studies have identified age- and gender-related biases in language models. Such findings matter for journalism because an AI system used to evaluate sources, summarize interviews or characterize individuals could reproduce those prejudices in news coverage. 

Consequently, responsible AI journalism requires more than proofreading. News organizations need to verify factual assertions against primary sources, review generated quotations and statistics, disclose meaningful uses of automation and maintain human editorial accountability. 

A journalist using AI to summarize a regulatory filing should still examine the underlying document. An editor publishing AI-assisted market commentary should verify the financial data and distinguish reported facts from speculation. 

AI can assist in gathering and presenting information fairly, but fairness is not guaranteed by automation. 

Can the Public Trust AI as a News Source? 

The public's skepticism toward AI-generated journalism is substantial, although its attitudes are more nuanced than simple rejection. 

A Pew Research Center analysis published in April 2025 found that 50% of American adults expected artificial intelligence to have a negative effect on the news people receive over the following two decades, compared with just 10% anticipating a positive effect. 

The same research found that 59% expected AI to reduce journalism employment. Asked to compare AI with professional journalists, 41% believed AI would do a worse job writing news stories, while 19% expected better performance and 20% anticipated roughly equivalent quality. 

Concern about inaccurate AI-generated information was particularly widespread, with 66% expressing extreme or substantial concern. Interestingly, Republicans and Democrats reported broadly similar concerns about AI's potential effects on news despite their different attitudes toward traditional media institutions. 

Public distrust, however, does not necessarily establish that AI-generated journalism is inferior. Research into how audiences perceive synthetic news has produced mixed findings, partly because readers evaluate the apparent credibility of the content differently from the credibility of its disclosed author. 

A 2026 meta-analysis of 31 studies examined how identifying news as AI-generated rather than human-written affects credibility judgments. The broader research raises an important distinction: The quality of a story and the reputation of its producer are related but separate questions.  

A well-researched, accurately sourced article does not become factually incorrect merely because AI assisted in writing it. Conversely, a human byline does not guarantee truthfulness, impartiality or competence. 

The more meaningful question is whether readers can trace information to credible sources, distinguish factual reporting from interpretation and determine who is responsible when something goes wrong. 

AI-generated news also introduces a less obvious danger: overdependence. 

A June 2026 study from the Massachusetts Institute of Technology Media Lab followed 67 participants over four weeks as they evaluated news items. AI assistance improved their immediate ability to detect fake news by 21%, but their unassisted performance subsequently declined by 15 percentage points relative to their initial performance. 

The findings suggest that AI can simultaneously improve a user's immediate judgment and weaken the independent skills needed to evaluate information without technological assistance. The researchers described this as an AI dependency paradox.  

For journalism, that distinction is fundamental. A technology that provides accurate answers most of the time may still leave its users more vulnerable when it makes mistakes. 

AI-Generated News Versus Social Media Influencers 

Comparisons between AI-generated news and professional journalism sometimes overlook another major competitor for public attention: social media influencers. 

Influencers operating on YouTube, TikTok, Instagram, X and other platforms increasingly function as news sources, commentators and interpreters of current events. Their audiences may regard them as more authentic, accessible and trustworthy than traditional news organizations. 

Yet influencer-generated news frequently operates without the institutional safeguards associated with professional journalism. There may be no independent fact-checking, editorial review, formal corrections process or meaningful separation between reporting and advertising. 

Influencers also face commercial incentives to maximize engagement. Controversial opinions, dramatic predictions and emotionally charged presentations can generate more attention than careful explanations of complicated issues. 

AI-generated news has different strengths and weaknesses. A well-designed system can compare multiple sources, summarize competing perspectives, identify inconsistencies and provide relatively dispassionate explanations. Unlike a human influencer, it does not personally benefit from cultivating celebrity or social status. 

Nevertheless, AI services can have their own commercial incentives, including user engagement, subscription growth and preferential treatment of particular information sources. Their outputs can also reflect training-data biases or the preferences embedded in system instructions. 

An influencer may distort a story to attract followers, while an AI system may distort the same story because its source material is incomplete or because it incorrectly synthesizes conflicting accounts. 

Both can produce confident, persuasive misinformation. 

There is no universal evidence-based ranking that establishes AI-generated news as more accurate than influencer journalism across every subject and platform. Accuracy depends on the particular creator or model, its sources, the reporting process and the degree of independent verification. 

The relevant distinction is not necessarily human versus machine. It is accountable, evidence-based information production versus unsupported assertion. 

The Financial Advice Analogy 

The comparison becomes especially important when applied to financial advice. 

Financial influencers, commonly called finfluencers, provide investment commentary, budgeting recommendations, retirement guidance and market predictions to enormous online audiences. Some offer valuable educational material, but others promote speculative investments, conceal financial relationships or make recommendations without adequately considering individual circumstances. 

Generative AI systems can offer an alternative that is more structured, patient and comprehensive. A chatbot can explain portfolio diversification, compare investment strategies, model hypothetical retirement outcomes and identify financial concepts that an inexperienced investor might otherwise misunderstand. 

It can also ask clarifying questions and present alternative scenarios without the emotional intensity often associated with social media investment commentary. 

However, AI-generated financial advice introduces its own problems. Models can rely on outdated information, invent financial statistics, misunderstand tax regulations or recommend unsuitable strategies because they lack sufficient knowledge of a user's circumstances. 

Moreover, neither an AI chatbot nor a social media influencer automatically owes the user the fiduciary duties associated with a regulated advisory relationship. 

Consider an investor seeking advice about whether to purchase shares in a highly valued AI company. An influencer might enthusiastically promote the stock based on recent performance, personal holdings or expectations of rapid growth. A chatbot might offer a more balanced discussion of valuation, earnings, competition and concentration risk. 

But the chatbot could also generate an inaccurate valuation, overlook material developments or implicitly favor the company's narrative because its training materials disproportionately reflect promotional coverage. 

The influencer's bias may be easier to recognize because it is expressed through enthusiasm, personality or commercial sponsorship. The AI system's bias may be harder to identify because it is embedded in an apparently objective explanation. 

For financial professionals, this suggests a role for AI as an analytical assistant rather than an unaccountable substitute for professional judgment. AI can help clients understand investment concepts and compare alternatives, but individualized recommendations still require accurate information, appropriate oversight and consideration of the client's goals, financial condition and risk tolerance. 

A well-governed AI financial tool may outperform a poorly informed influencer. A knowledgeable, ethical human advisor may outperform both. The comparison should be based on demonstrated accuracy, suitability and accountability rather than assumptions about which type of communicator is inherently more trustworthy. 

Is Artificial Intelligence More Biased Than a Human Journalist? 

Determining whether AI is more biased than the average newsroom writer or editor requires defining what bias actually means. 

Human journalists possess personal experiences, political beliefs, professional relationships and cultural assumptions that influence their interpretation of events. Even conscientious reporters make subjective decisions about which stories deserve attention, which sources are credible and which facts are most relevant. 

News organizations also have institutional biases. Their editorial priorities may reflect audience demographics, ownership structures, commercial relationships or longstanding ideological commitments. 

AI systems inherit different forms of bias. Their training data reflects historical inequalities, editorial judgments and the relative visibility of different perspectives. Their developers make choices about which information to include, how models should respond to controversial subjects and what constitutes an acceptable answer. 

Researchers have demonstrated that these problems extend beyond overt political opinions. A Stanford University study published in 2025 identified age- and gender-related distortions in generative AI outputs, including the tendency to portray working women as younger and less experienced than comparable men.  

Yet AI has certain advantages over human journalists. It can be systematically tested using standardized questions, evaluated across thousands of responses and compared against measurable benchmarks. Developers can modify training procedures and instructions to reduce identified problems. 

Human bias is often more difficult to measure and correct. A journalist may sincerely believe that personal judgments are objective, while an editor may unconsciously favor familiar narratives. 

Nevertheless, AI's apparent neutrality can be misleading. A chatbot may produce balanced language while omitting important facts or presenting unsupported perspectives as equally credible. 

A human reporter can also exercise independent judgment in ways that AI systems cannot reliably replicate. Journalists can interview witnesses, investigate physical evidence, challenge institutional authority and accept professional responsibility for their reporting. 

The answer, therefore, is not that AI is inherently more or less biased than human journalism. Different forms of bias emerge under different conditions, and their consequences depend on how information is collected, evaluated and published. 

The most reliable approach combines machine-assisted analysis with human editorial oversight, independent verification and transparency about the limitations of both. 

When Media Bias and AI Bias Reinforce Each Other 

Perhaps the most consequential relationship between artificial intelligence and journalism is the possibility that their biases will reinforce one another. 

Large language models learn from enormous collections of human-generated material, including journalism, commentary, websites and other published information. Consequently, the choices made by news organizations can influence the information that AI systems subsequently reproduce. 

If media coverage consistently emphasizes certain perspectives while neglecting others, those imbalances may become embedded in AI-generated explanations. 

The reverse process is equally important. As newsrooms increasingly use AI for research, summarization, editing and content production, the assumptions embedded in models can influence what journalists publish. 

This creates a potential feedback loop. 

A news organization publishes a story with a particular interpretation of an event. An AI model incorporates that interpretation into its knowledge or retrieves it as supporting evidence. Another journalist uses the model to research the same subject, potentially reproducing the original interpretation without recognizing its limitations. 

Eventually, repeated exposure can make a particular narrative appear more authoritative simply because it has been reproduced across numerous platforms. 

The danger increases when AI-generated material itself becomes part of future training datasets. Errors and distortions can circulate between automated systems and human publishers, making it increasingly difficult to distinguish independently verified information from recycled assertions. 

For financial markets, such feedback loops could be especially consequential. An exaggerated narrative about an emerging technology, economic trend or publicly traded company could spread through news articles, AI-generated summaries, social media commentary and automated investment research. 

Investors might encounter apparently independent sources that ultimately derive from the same original claim. 

This does not mean that AI-generated financial news will necessarily manipulate markets. It means that information provenance—the ability to identify where claims originated and how they were verified—becomes increasingly important. 

Financial institutions should be particularly cautious when multiple AI systems produce similar conclusions. Agreement between models does not necessarily constitute independent confirmation when those models rely on overlapping training data or information sources. 

Can AI Detect and Correct Media Bias? 

Artificial intelligence may also provide one of the most promising tools for identifying bias in journalism. 

Researchers at the University of Pennsylvania have developed a Media Bias Detector that uses large language models to examine how news organizations select, frame and present information. 

In October 2026, researchers associated with Penn's Computational Social Science Lab described an expanded framework for analyzing media coverage. Their work included a dataset of more than 140,000 articles published during the 2024 presidential election year. 

Rather than assigning a single ideological label to each publication, the system examines dimensions including political orientation, tone, subject selection, quotations and differences in coverage of the same events. 

The researchers found, among other things, that news organizations devoted more than four times as much coverage to the presidential horse race as to policy topics combined. They also identified systematic differences between the political orientation of headlines and the articles accompanying them.  

These findings illustrate an important distinction between factual accuracy and editorial bias. A news article can contain entirely accurate statements while presenting an incomplete or distorted picture through its choice of subject matter, sources or emphasis. 

AI can help reveal these patterns at a scale that would be impractical for human researchers working alone. 

For example, an automated system could compare how dozens of publications describe the same corporate earnings announcement, identify emotionally charged language and examine whether positive and negative developments receive proportionate attention. 

It could also analyze whether financial reporting consistently favors large technology companies over smaller competitors, whether particular market risks are systematically underreported or whether coverage of AI investment disproportionately emphasizes either speculative enthusiasm or catastrophic predictions. 

However, automated bias detection must itself be treated skeptically. An algorithm cannot determine what constitutes fair coverage without making assumptions about relevance, proportionality and acceptable framing. 

Penn researchers acknowledge this limitation, emphasizing the measurement of observable differences rather than claiming to identify a universally correct editorial perspective. 

The Reuters Institute for the Study of Journalism has similarly examined the practical difficulties of detecting and mitigating AI bias in newsrooms. Its August 2025 analysis highlighted problems involving translation, training-data representation, recommendation algorithms and cultural assumptions, while describing how news organizations are experimenting with safeguards.  

For financial services organizations, comparable tools could strengthen compliance monitoring, investment research and communications oversight. AI could identify potentially misleading investment commentary, compare financial disclosures against published interpretations and flag language that exaggerates either risks or expected returns. 

Yet these applications should supplement, rather than replace, professional review. 

Toward a More Accountable Information Ecosystem 

The debate over artificial intelligence and media bias ultimately exposes a problem that predates generative AI: Neither human judgment nor technological sophistication guarantees objectivity. 

Traditional journalism has always struggled with political pressures, commercial incentives, incomplete information and individual prejudice. Artificial intelligence inherits many of those weaknesses while introducing new problems involving opaque algorithms, fabricated information and the rapid reproduction of errors. 

At the same time, AI offers capabilities that could improve journalism considerably. It can help reporters analyze complex evidence, expose inconsistencies, compare competing narratives and investigate institutional behavior at unprecedented scale. 

The appropriate goal is not to eliminate human judgment from journalism or to reject AI because its outputs are imperfect. It is to create systems in which both human and machine-generated information are subject to meaningful scrutiny. 

For media organizations, that means maintaining editorial accountability, disclosing consequential AI use, verifying information independently and resisting the temptation to sacrifice accuracy for speed or cost savings. 

For AI developers, it means improving source transparency, testing for systematic bias and providing mechanisms for correcting inaccurate or misleading outputs. 

For financial professionals, it means treating AI-generated information as an input to analysis rather than an unquestionable authority, particularly when investment recommendations or client decisions are involved. 

And for the public, it means developing the critical skills necessary to evaluate information regardless of whether it comes from a journalist, influencer or chatbot. 

Artificial intelligence does not have to receive favorable coverage to receive fair coverage. Journalism does not have to abandon skepticism to acknowledge technological progress. And the public does not have to choose between trusting humans and trusting machines. 

The more productive standard is to demand evidence, transparency and accountability from both. 

Ultimately, the greatest danger may not be that artificial intelligence is biased against journalism or that journalism is biased against artificial intelligence. It is that each could validate and amplify the other's mistakes, creating an information environment in which repetition is mistaken for verification and confidence is mistaken for truth. 

The opportunity is to use the strengths of each to expose the weaknesses of the other. That would represent meaningful progress not only for journalism and artificial intelligence, but also for the financial markets and institutions that depend on trustworthy information.    

Researched by DWN Staff