7 Ways to Spot AI Writing

Here are some tips for determining if an article is likely written by AI.

73. OVERUSED WORDS. AI-written articles tend to come back to the same terms multiple times. Examples would be comprehensive, delve, meticulous, versatile and pivotal. Before 2024, overused AI words in scientific research papers were typically nouns. More recently, researchers say AI excessively uses "style" words—mostly verbs and some adjectives.

74. TORTURED ACRONYMS. Generative AI will sometimes pick up the wrong words for an acronym. For instance, a data science paper might use "CNN" to refer to "convolutional brain organization" instead of "convolutional neural network.”

75. NONSENSICAL PARAPHRASES. An academic paper written by AI might have “glucose bigotry” instead of “glucose intolerance,” where it changed a single word and did not recognize the context.  

76. ACADEMIC CITATIONS. AI-written articles with academic citations have been known to include incorrect or incomplete references. 

77. STYLE CHANGES. A sudden change in writing style within an article or essay may indicate that the author’s work was rewritten using AI.

78. PERFECT GRAMMAR. A typo, particularly in student writing, could indicate the article or essay is not wholly the work of a bot. Mistake-free writing is, ironically, a red flag. However, savvy writing prompts may ask the AI to include some errors in order to mislead inspectors.  

80. MECHANICAL STYLING. AI tends to mechanically repeat expressions that appear often in the internet material that it was trained on. The result is often uninspired and generic prose that often lacks any specific point. 

 More about spotting fake news

AI Definitions: Machine Learning

Machine learning (ML) - This subset of AI makes predictions or decisions based on patterns it spots in data sets. The process evolves and adapts on its own as it is exposed to new data, improving the output without explicit programming from a human. An example would be algorithms recommending ads for users, which become more tailored the longer it observes the users‘ habits (someone’s clicks, likes, time spent, etc.). Data scientists combine ML with other disciplines (like big data analytics and cloud computing) to solve real-world problems. However, the results are limited to probabilities, not absolutes. It doesn’t reveal causation. There are four types of machine learning: supervised, unsupervised, semi-supervised, and reinforcement learning. A clever computer program that simply mimics human-like behavior can be considered AI, but the computer system itself is not machine learning unless its parameters are automatically informed by data without human intervention. Video: Introduction to Machine Learning

More definitions here

Bent to the Illusion

The Müller-Lyer illusion, a famous optical illusion (involves) two sets of arrows. The arrows are exactly the same length. But in one case, the ends of the arrows outward, seem to signify expansion and boundless potential. In the other case, they point inward, making them seem self-contained and limited. The first case is analogous to how investors see the stock market when returns have been increasing; the second case is how they see it after a crash.

“There’s no way that you can control yourself not to have that illusion,” (Nobel prize winner) Daniel Kahneman told me. “You look at them, and one of the arrows is going to look longer than the other. But you can train yourself to recognize that this is a pattern that causes an illusion, and in that situation, I can’t trust my impressions; I’ve got to use a ruler.”

Nate Silver, The Signal and the Noise

4 Fake News Signals from Outside the Website

Here are some tips for determining if a story is likely reliable. An organization does not need to tick off all these qualifiers in order to be considered authentic and accurate, but the more you see red flags pop up, the more a healthy skepticism is in order.

69. YOUR COMMUNITY. There’s no substitute for knowing people who are well-informed and will let you know when you’ve posted something questionable. These are people you can ask when you have your doubts. Don’t know any experts or researchers, or information junkies from various fields who are critical and helpful? Make some new friends! Developing such a support system is critical for navigating effectively through life. Read some books written by experts. 

70. FACT-CHECKING SITES. Does a fact-checking site identify the assertion of the article as a hoax? Check one of the sites listed at the end of this article or type the article’s topic into a search engine and add “hoax” or “fake.”  

71. THE OTHER SIDE. Take time to check sites that do not agree with your politics. If you discover they are wrong and perhaps not addressing the best arguments of your side, it is a confirmation you are on the right side of an issue. Or maybe you will discover a weakness in your own reasoning you haven’t considered. Either way,  you'll know what other people are consuming, sharpening your thinking.

72. GOOGLING THE TOPIC. If you do a Google search for a topic, remember that reliable researchers do not write material answering questions like “Did the Holocaust exist?” Instead of decent sources, this type of search will bring up conspiracy theorists. Don’t be misled by a search that frames issues as secret plots and nefarious schemes.

 More ways to spot fake news

AI Definitions: Hallucinations

Hallucinations – This is when an AI provides responses that are inaccurate or not based on facts. Generative AI models are designed to generate data that is realistic or distributionally equivalent to the training data and yet different from the actual data used for training. This is why they are better at brainstorming than reflecting the real world and why they should not be treated as sources of truth or factual knowledge. Generative AI models can answer some questions correctly, but this is not what they are designed and trained to do.

More AI definitions here.

12 Fake News Signals from the Publisher

Here are some tips for determining if a story is likely reliable. An organization does not need to tick off all these qualifiers in order to be considered authentic and accurate, but the more you see red flags pop up, the more a healthy skepticism is in order.

57. REPUTATION. Is the writer’s reputation at stake if they are wrong? Does the organization risk losing reputation or losing finances if it becomes known for having promoted false news? 

58. RELIABILITY. Has the organization been reliable in the past? Have you read other information from the organization confirmed to be accurate?

59. AMATEURISH. Data collected by an amateur is more error-prone than data collected by a professional scientist. Does a quick web search confirm whether the people who collected and organized the data have a good track record of collecting and distributing data?  

60. RESPONSE TO CRITICS. Does the publisher respond publicly to its critics when there are reasonable questions? Does it acknowledge when the critics have a point? 

61. DATA SOURCES. Look closely at the sources of data the publisher uses: is this material provided by for-profit companies, partisan organizations, or advocacy groups? While the material may be accurate, data from groups with agendas require greater scrutiny than data from nonpartisan organizations. 

62. PAYING THE WRITERS. Content Farms (or Content Mills, if you like) pay very little in return for lots of writing. When news writers are focused on cranking out material to feed the beast, the quality of the work suffers. If you discover a site is considered a Content Farm by professionals or pays writers very little for their work, that’s a big red flag. 

63. DIVERSE VOICES. Does the news organization offer diverse perspectives in its articles? A professional outlet will make a concerted effort to give voice to various ethnicities and political persuasions. The more a newsroom focuses on a single viewpoint, the greater the likelihood it will leave out significant perspectives from its news converge.

64. FEEDBACK. Reputable news publishers want readers’ feedback on stories for accuracy and look for help in determining coverage priorities.  

65. AGREEMENT. Do you find yourself agreeing with everything your preferred news outlet says? If so, something is wrong. Find a commentator whose politics don’t match with your own—vary your media consumption to get a balance of perspectives. 

66. EASY STORIES. Suppose a news outlet overlooks stories worth telling in favor of the stories that can be easily told. In that case, it may not have the resources to dive into investigative reporting or may not have the goal of getting beyond the low-hanging fruit.

67. ANONYMOUS SOURCES. Legitimate news outlets will only reference unnamed sources that would endanger them physically or put them in legal jeopardy. Overreliance on anonymous sources should be a red flag to be skeptical of the information, even if it comes from an otherwise trustworthy site.   

68. FRAMEWORK. Some sites have a framework for all their stories (like the College Fix, which is focused on college campus outrage). Articles on these sites may leave out moderating information, so stories lean toward the framework.

 More ways to spot fake news

Enduring mistreatment to justify revenge

A woman will seek psychiatric attention for depression in response to desertion by her husband. She will regale the psychiatrist with an endless tales of repeated mistreatment by her husband.

The therapist discovers that this pattern of mistreatment has existed for twenty years, and that while the poor woman divorced the brute of a husband twice, she also remarried him twice, and that innumerable separations were followed by innumerable reconciliations.

What is going on here? In trying to understand what has happened, the therapists recalls the obvious relish with which the woman had recounted the long history of her husband’s brutality and mistreatment. Suddenly a strange idea begins to dawn: maybe this woman endures her husband’s mistreatment, and even seeks it out, for the very pleasure of talking about it. But what would be the nature of such pleasure? The therapist remembers the woman’s self-righteousness. Could it be that the most important thing in the woman’s life is t have a sense of moral superiority and that in order to maintain this sense she needs to be mistreated? The nature of the pattern now becomes clear. By allowing herself to be treated basely she can feel superior.

If the world is treating us well we have no need to avenge ourselves on it. If seeking revenge is our goal in life, we will have to see to it that the world treats us badly in order to justify our goal. Masochists look on their submission to mistreatment as love, whereas, in fact it is a necessity in their never-ceasing search for revenge and is basically motivated by hatred.

M Scott Peck, The Road Less Traveled

11 Webinars This Week about AI, Media, Journalism & More

Tue, Aug 6 - Why Can’t I Just Google It?  

What: Join us to lift the curtain and explore the motivations and methods behind this ubiquitous search engine and learn how CloudSource can open the door to 60 million open access and open educational resources.

Who: Maryska Connolly CloudSource, Director of Partnerships & Communications  

When: 1 pm, Eastern

Where: Zoom

Cost: Free

Sponsor: CloudSource

More Info

 

Tue, Aug 6 - From Boomers to Zoomers: Cracking the Code of Cross-Generational Engagement 

What: In this session, participants will: Understand the needs and concerns of each generation, gain skills and strategies to engage and train each generation, and learn how immersive and game-based learning increases participation and knowledge retention.

Who: Ellen Burns-Johnson, VP of Learning Strategy - Growth & Strategy Team, ELB Learning 

When: 3 pm, Pacific

Where: Zoom

Cost: Free

Sponsor: ELB Learning 

More Info

 

Tue, Aug 6 - What Journalists Should Know About Election Security: Threats and Mitigations

What: Attendees will learn about: the landscape of threats to the 2024 U.S. elections, what those threats mean practically to the security or integrity of the elections process, what election officials, vendors and the federal government are doing to protect the security and resilience of the U.S. election infrastructure.

Who: Cait Conley is the Senior Advisor to the Director, a role that includes responsibilities supporting CISA’s election security efforts. Conley leads CISA’s work in partnering with state and local election officials to manage and reduce risk to the Nation’s election infrastructure.

When: 12:30 pm, Eastern

Where: Zoom

Cost: Free

Sponsor: Harvard’s Shorenstein Center on Media, Politics and Public Policy

More Info

 

Tue, Aug 6 - The Formula for Social Media Success

What: Learn the key differences between social networks Identify your target market. Set your social media goals Build your content strategy. Create your Ad Strategy Measure your results. Discover must-have social media tools.

Who: Ray-Sidney Smith, Digital Marketing Strategist, Hootsuite Global Brand Ambassador, Google Small Business Advisor for Productivity, and Managing Director of W3C Web Services.

When: 10 am, Eastern

Where: Zoom

Cost: $45

Sponsor: Pennsylvania Business One-Stop Shop

More Info

 

Tue, Aug 6 - Solutions Journalism 101 Webinar

What: This webinar will explore the ins and outs of solutions journalism, talk about why it’s important, explain key steps in reporting a solutions story, and share tips and resources for journalists interested in investigating how people are responding to social problems. We will also explore additional resources we have on hand for your reporting, including the Solutions Story Tracker, a database of more than 16,000 stories tagged by beat, publication, author, location, and more, a virtual heat map of what’s working around the world.

When: 6 pm, Eastern

Where: Zoom

Cost: Free

Sponsor: Solutions Journalism Network

More Info

 

Wed, Aug 7 - The Role of Visual Content for Journalists and PR Professionals

What: Learn about journalism’s transformation over time and the impact on PR strategies. Video content and how it enhances media coverage Insights about how journalists research the stories they cover.

Who: Peter Banda Assistant News Director, US News Gathering, The Associated Press; Anthony Brito Senior Director, Content Services, The Associated Press

When: 1 pm, Eastern

Where: Zoom

Cost: Free

Sponsor: Associated Press, Business Wire

More Info

 

Wed, Aug 7 - How to start and grow a newsletter

What: In this session, a few expert newsletter writers and I will dive into how to build a successful newsletter, from launching and finding an audience to growing, monetizing, and maintaining it.

Who: Tim Herrera, author and former editor at The New York Times's Smarter Living; Lyz Lenz of the newsletter Men Yell At Me; and Parker Molloy of The Present Age.

When: 4 pm, Eastern

Where: Zoom

Cost: $20

Sponsor: Freelancing with Tim

More Info

 

Wed, Aug 7 - How to Explain Data Through Visualization and Storytelling      

What: You’ll learn:  Data visualization basics to help get you started. The tools and technologies that promote more effective data use. How successful organizations have made data storytelling a reality.

Who: Trang Tran, Deputy Chief Data Officer, Director Artificial Intelligence COI, Office of Information and Technology, U.S. Customs and Border Protection; Prakhar Bajpai, Senior Data Scientist, City of Austin & 2023 NextGen Public Service Award Winner

When: 2 pm, Eastern

Where: Zoom

Cost: Free

Sponsor: GovLoop

More Info

 

Thu, Aug 8 - AI + Data Journalism

What: In this participatory practical training session, we’ll work with Data Analyst, Julius.AI, Claude and others to analyze and visualize data. Participants will analyze a dataset of bridge inspection records using these tools. We’ll work with data scrape formulas and tools such as Tabula and PDFtoExcel.com as well.

Who: Mike Reilley Senior Lecturer, University of Illinois-Chicago

When: 2 pm, Eastern

Where: Zoom

Cost: Free to members, $25 for nonmembers

Sponsor: Online News Association

More Info

 

Thu, Aug 8 - Unleashing the Power of Collaboration: A Toolkit for Community Newsrooms

What: The results of in-depth interviews on collaborative journalism invovling 39 outlets, 44% serving local audiences, and how you can develop collaborations that serve your community.

Who: Bridget Thoreson, Chief Project Officer/Dream Wrangler at the social impact firm Hearken

When: 2 pm, Eastern

Where: Zoom

Cost: $35

Sponsor: Online Media Campus

More Info

 

Sun, Aug 11 - How to write better personal essays

What: Not sure how to turn a personal story into an essay you can sell? Wondering how to structure a personal essay? How do you even pitch one? We’ll cover all those questions and more in this workshop.

Who:  Tim Herrera, author and former editor at The New York Times's Smarter Living; Expert essayist and editor Matt Ortile.

When: 12 pm, Eastern

Where: Zoom

Cost: $20

Sponsor: Freelancing with Tim

More Info

12 Fake News Signals from the Website

Here are some tips for determining if a story is likely reliable. An organization does not need to tick off all these qualifiers in order to be considered authentic and accurate, but the more you see red flags pop up, the more a healthy skepticism is in order.

45. ABOUT US. Check the site’s About page for information about who is behind the operation. If you aren’t familiar with the name, look for information about who owns it. For instance, the Russian government owns the RT network. There is a big difference between state media (RT) and public media in a democracy (like the BBC). If a website does not provide information on its mission, staff members, or physical location, it is most likely unreliable. The language used here should be straightforward. If it seems overblown, be skeptical. 

46. ADDRESSES. There should be a mailing address (better yet, a physical address) and an email address. Any site concerned about making factual corrections (and avoiding defamation) needs a way for readers to contact them.  

47. LEGAL NOTICES. Look for a legal section on the website. It may be called a “disclaimer.” Satirical websites sometimes disclose this information in those sections. A site without obvious legal notices (such as EEOC or FCC public file information for TV stations) is a red flag.  

48. GOOGLE “FAKE.” Put the website name in quotes and then add “fake.” Something indicating the site is known for publishing fake news might come up.

49. DATES. Look for a date to make sure the event is recent. Sometimes real stories from several years ago are posted as if they were new. This happens with photos as well. Reliable news outlets want readers to know when the information is posted and will usually display the headline's date. If you are looking at an article on social media, go to the article and look first for a timestamp. Even an old article with good information at the time of publication can be problematic because a study (for instance) may have since been retracted.

50. WEB DESIGN. Poor web design is a red flag. Is the design out of date when compared to other reputable sites? Is the display navigable and professional? 

51. DOWNLOADS. If the website contains advertisements, particularly pop-up ads, asking you to download software, it is likely to be an unreliable.

52. CORRECTIONS. Does the site make corrections as it receives new information, and does it make those corrections obvious? Typically, a note will be added to the top or bottom of a news article when even a factual change is made to a story. In a print or broadcast story, the original error should be clearly stated along with the correct information. The editorial process of a legitimate news organization catches and corrects many errors. If you don’t see corrections from time-to-time on a website, that’s a red flag. Corrections and updates are a part of journalism.  

53. OTHER ARTICLES. Search for the information you know to be false in other articles on the site. Does the site offer quality information on different topics besides the one you are investigating?

54. COMMUNITY POSTS. Some sites allow individuals to post pieces under the banner of the news brand (ex: BuzzFeed Community Posts, Kinja blogs, Forbes blogs). The site editors typically do not vet these posts, making the material suspect.

55. PREVIOUS FAKE NEWS. Do Wikipedia, Snopes, or other such sites show the website in question as having a connection to spreading false information in the past? While Wikipedia is generally pointed in the right direction but can contain some questionable information, the links to other sites it provides can be invaluable in the hunt for the truth.   

56. EMBARGOS. Does the publisher respect embargos? This is common practice in media, where information suppliers ask publishers to hold back new information until a certain time. It is considered common courtesy and accepted practice to honor embargos except in unusual circumstances. Ignoring these expectations could be a sign the publisher is more interested in rushing out material than operating by industry standards.

 More ways to spot fake news

Fake News Signals from Website Addresses

Here are some tips for determining if a story is likely reliable. An organization does not need to tick off all these qualifiers in order to be considered authentic and accurate, but the more you see red flags pop up, the more a healthy skepticism is in order.

38. DOT-GOV. Does the site have a dot-gov address? Generally, data provided by government organizations are more trustworthy. Official news agencies are often the starting place for establishing the truth of a matter.   

39. DOT-EDU. Does the site have a dot-edu address? Generally, data provided by university research laboratories are reliable.

40. LO & DOT-CO. Websites ending with odd letters like “lo” (such as “Newslo”) or “.co” could be a red flag for fake news sites. For instance, abcnews.com.co is not the actual URL (web address) for the real ABC News. What looks like an .edu domain, followed by .co or “lo” is likely a fake or deceptive site. Sites like Clone Zone make this easy to do.

41. COM.COM.  Another way to trick readers is to add a “com” so the web address almost looks right. For instance, “USATodaycom.com.”

42. PERSONAL BLOGS. A domain such as “.wordpress” or “blogger” usually signifies it’s a personal blog rather than a news source.

43. COUNTRY-CODE TOP-LEVEL DOMAINS. The end of some website addresses is a clue to the site’s originating country. For instance, “dot-au” means the site is based in Australia, and “dot-ng” means the site is based in Nigeria. Suppose you find a supposed article about your community on a website coming from a country far away. In that case, it probably means the writer isn’t likely to have access to the necessary sources to write a competent story. 

44. ODD NAMES. Odd domain names generally equal odd and rarely truthful news.

More on spotting fake news

How long does it take to Read a Person’s Emotions?

Psychology professor William Ickes found that our ability to read strangers during a six-minute interaction is limited, but other experiments showed that we have much greater accuracy when reading our friends’ behavior. That’s not really shocking, but what is surprising is that when study subjects spent just a little more time with a stranger, their ability to read that person was just as good as their ability to read close friends. The magic time span? Thirty minutes. Time, or at least the amount of information acquired over time, is crucial to empathy. That is not only the case for real-life encounters, but also for stories.

Lene Bech Sillesen writing in the Columbia Journalism Review

14 Fake News Signals around Articles

Here are some tips for determining if a story is likely reliable. An organization does not need to tick off all these qualifiers in order to be considered authentic and accurate, but the more you see red flags pop up, the more a healthy skepticism is in order.

24. OUTRAGE HEADLINES. Fake news outlets have found stories that make the reader angry can generate more shares. The use of ALL CAPS or numerous exclamation points is a red flag. If the headline is compelling, then read the entire story, not just the headline. Does the story deliver what the headline promises?

25. HYPERBOLIC HEADLINES. Hyperbolic headlines claiming you’ll “never believe” the article’s epic content suggest you shouldn’t click. If the headline promises to tell you something the "media doesn't want you to know," as if you are being given secrets to an inside story, be skeptical. 

26. AUTHOR ATTRIBUTION. Lack of author attribution can mean the news story is suspect. Some respected sites, such as The Economist, don’t typically identify their  writers, but that’s an exception to the rule. Nearly all quality news outlets will identify the writer(s) of each article. 

27. AUTHOR CREDENTIALS. Look for other articles by the same author by Googling the person’s name. Have they produced legitimate writing for legitimate news outlets? Their credentials and backgrounds are a guide to the quality of work they are likely to produce as well as the quality of the news outlet you are considering. Suppose the story is about a specialized area, such as health or science. In that case, it’s a bonus if the author regularly writes about the subject because the person is also likely to possess basic knowledge of that particular area. 

28. SUPPORTING PHOTOS. Do accompanying photos visually back up the story’s claims? Do the images even relate to the headline and content? If a site uses lots of stock photos rather than original images related to the story, that’s a red flag—perhaps that the reporters are not professional, but it could indicate more serious problems.  

29. ODD PHOTOS. Are photos cropped oddly or taken from some strange angle? Does it appear the photographer was deliberately trying to avoid showing something in the shot? Legit news organizations avoid picking unflattering photos that might bias the reader unless that’s part of the story. This goes for distorted images as well, taken very close to a subject to emphasize someone’s age or physical characteristics. When a site shows a politician or celebrity’s face contorted or just plain goofy in a photo, it’s a subtle attempt to affirm your negative impression of that person and cue you that the article will fit your bias.   

30. MANIPULATED IMAGES. Sometime real images are cropped to give viewers a misleading impression. One way to check a photo is to use Google Reverse Image Search. Paste in the image link or upload the photo. This should give you information on where else the image has been posted so you can compare for manipulation. This will also show if there is mislabeling, and the photo has been repurposed from another event or time. Other free reverse image search engines include TinEye and RevEye. Another way to check graphics is to match the image against Google Street View or satellite imagery of the location. Consider details like vehicles and weather for consistency.

31. MISLEADING GRAPHS. Look closely at any charts related to an article. Using plot points that misinterpret data can skew the results displayed in the image. The axis should always have labels and the value should start at zero. A pie chart should not add up to more than 100%. Trends over significant spans of time are more meaningful than isolated events.

32. GRAPH INFORMATION. Legitimate news sites will include the source of information from which infographics are built. A graphic that does not include a data source is a major red flag; the information is either false or the organization fails to maintain ethical and professional practices. The source of the data can be checked for more information on the topic. There better be a good reason to see a three-dimensional chart. They are seldom needed.   

33. BAIT & SWITCH. Does the headline match the article? Many compelling headlines don't. Reliable websites respect readers by avoiding discrepancies between the headline and the story. Teases designed to trick readers into clicking are a sure sign of a disreputable organization. Reputable sites deliver on the headline’s promise and do not frustrate readers by holding back information in the headline.

34. MEDICAL STUDIES. If the article is based on a new study, the “gold standard” are double-blind peer-reviewed studies. Double-blind studies mean that neither the researchers nor the subjects know who is getting the real medicine or a placebo. If the study is a preprint, it has not been peer-reviewed. If the study was conducted on animals, it does not necessarily mean the findings will apply to humans. Beware of studies from so-called “predatory journals.” These publications do not peer-review manuscripts they publish, and they charge authors a fee to publish.  

35. CONTENT TAGS. Some social media platforms are trying to counter misinformation by adding tags to content that has been identified as misleading. 

36. EDITED VIDEO. If video accompanies an article, look for multiple edits and odd cuts. Amateurs with basic software can easily create cheap fakes by making edits that slow down video, speed it up, cut it into snippets, insert or remove details, or present it in a false context. 

37. FAKE VIDEO. The development of “deep fakes”—videos created using artificial intelligence is making it more difficult to identify manipulated images. Technology can swap faces, clone voices and synchronize lips to a different audio track than the original. A screenshot from a questionable video placed in Google Reverse Image Search might show if there history of manipulation for the image and thus the video as well. Look for visual cues that suggest the video is manipulated, such as distortions, odd movements, and syncing issues between the voice and the facial movements.

More on spotting fake news

Becoming better persons

We cannot sharply separate the process by which we come to have true beliefs from the process by which we try to become better persons. If we demand intellectual certainty about our beliefs before we begin to live out those beliefs, we will not make much progress on the intellectual questions themselves... progress in answering our intellectual questions goes hand in hand with progress in becoming better people. Our beliefs matter... but we cannot hope to settle questions of belief in a way that is prior to and independent of our struggle to become selves of a certain sort.

C Stephen Evans, Kierkegaard: An Introduction

23 Fake News Signals from the Writing

Here are some tips for determining if a story is likely reliable. An organization does not need to tick off all these qualifiers in order to be considered authentic and accurate, but the more you see red flags pop up, the more a healthy skepticism is in order.

You’ll notice these are “tips” and not a checklist. Checklists can oversimply the nuances of discernment into black-and-white boxes.

1. ORIGINAL REPORTING. Does this article cite sources likely to know this information? Does the news organization have reporters attending news conferences in person, working in cities where the news is happening, and talking to key sources directly? Or does the organization have to rely on second-hand information from other sites?

2. LONE-WOLF REPORTING. Compare the information with other sites you trust. Are these sites reporting the same information? The site might have a scoop, but a lack of multiple independent accounts means it is more likely that the story is false. Sometimes lack of coverage is the result from writer and producer bias within a company or the result of the particular focus of the outlet (which may include not offending certain sponsors or other companies owned by the same parent company). Typically, you should expect more than one source reporting on an important topic or event. Plus, it’s always best to read multiple sources of information to get various viewpoints and media frames.

3. AP STYLE. Most legitimate news organizations will use the AP Stylebook as a writing guide (no Oxford comma, full name on the first reference and only last name thereafter, etc.). Some organizations have developed their own style guides (New York Times, Wall Street Journal, etc.). Most news organizations use an in-house style guide (to deal with writing issues unique to the publication’s area of reporting).   

4. POOR GRAMMAR. When a writer makes obvious grammatical mistakes, they also may not have taken the time to ensure the article’s facts are accurate. 

5. ADJECTIVES & ADVERBS. Objective journalism avoids adjectives and adverbs. The more of them used in an article, the more you should question whether the writer’s goal is to inform you or to convince you of something.

6. BALANCE. Did the writer engage with anyone who disagreed with the gist of the article? Quality news organizations are looking for both sides—and sometimes, there are more sides than that. 

7. EXPERTS. Does the article quote not only more than one side in a dispute but experts as well? A he-said-she-said story without experts’ opinions in the field is weak reporting. When there’s only a single source for a news article, be hesitant to accept the information without further corroboration.

8. OBJECTIVE. Like the scientist aiming to discover the truth, having some bias does not mean a journalist cannot arrive at the truth through a tested and effective approach (as does the scientific method, despite the researcher’s bias). The complaint that “no one can really be objective” misses the point that it’s not the journalists themselves but the articles that need to be neutral. While bias websites can still post real news, carefully look at the specific evidence, they offer and see if reporting from other legit sites backs it up.

9. OPINION. Is the article part of an opinion section? Does the video feature a commentator? Commentary has a long history of having a part inside the pages of newspapers, but many readers confuse an editorial article with news reporting. The same can happen online or on TV news. There’s no need for an opinion piece to be neutral in its presentation. Just don’t confuse it with an unbiased news piece.           

10. GENERIC ENEMIES. Does the article focus on vague foes who are never specified? “The media,” “supporters of (insert name of politician),” “The right,” “The left,” “Washington,” etc. Good reporting doesn't make these kinds of generalizations.

11. DOXING. Doxing is making private information public in order to hurt a person or organization. If the writer suggests anything like doxing, run the other way.

12. EMBEDDED LINKS. Quality journalism values clarity over style. Links in the article to original source material show a commitment to transparency and allow readers to make up their own minds about its use. Sometimes bogus stories will cite official or official-sounding sources and even link back to them that do not back up the claims in the article. An article without links or quotes from identified sources should be suspect.

13. SPONSORED CONTENT. Some legit news organizations publish articles similar to what they usually publish as real news—only, in this case, an advertiser actually sponsors the material. The intention could be to provide legitimate information about a subject while at the same time promoting the advertiser's product. Sometimes referred to as native advertising, reputable publishers will identify the article as “sponsored content” or “paid partner content” in a prominent location. A precursor to sponsored content was advertorials—a combination of advertising and editorial opinion. These placements were ads disguised as editorial content. 

14. LOCAL REPORTING. If the story involved a particular locale, was local expertise included? Was the reporting conducted on the scene?

15. YOUR REACTION. Be sensitive to occasions when you become angry as you read an article. If you are outraged after reading something, the story may be written to manipulate your emotions. The more shocking and outrageous, the more work is necessary to confirm the information before passing it along.

16. PARTISAN APPEAL. If a story sounds big but appears only on hyperpartisan sites and seems designed for outrage, it could have significant flaws that stopped legitimate news outlets from covering it.

17. FIRSTHAND SOURCES. Use an article’s information to work back to original sources to verify what’s in the article. If the report references a lawsuit, it can often be found online through a Google search or third party like Scribd, CourtListener, or DocumentCloud. Or, if the article references something a company is doing, check that company’s website (or a government agency) to see if there is a news release about that topic or an announcement on a site like PR Newswire.    

18. MULTIPLE SOURCES. Use keywords from the article (unique terms such as someone’s name) that are likely to bring up the same topic from another source using Google News search. The information from each story can be compared. It is unusual for a single outlet to have exclusive information, especially after several days have passed since the article was posted. To avoid generic, unhelpful search results, use unique keywords in your search—like the name of an unfamous person who’s quoted in the story. 

19. TANTALIZING QUOTES. Search for a questionable quote by pasting the text (in quotation marks) into a search engine. If the exact phrasing doesn’t come up or if only a few small outlets have printed the outrageous quote (perhaps from a famous person or politician), then be skeptical about its authenticity. Look for the sentences before and after the quote that makes your blood boil. If the article fails to give them, that could be a warning sign. If the quote is taken out of context, the site (or writer) belongs on your naughty list.

20. SOCIAL MOVEMENTS. A social movement or political uproar can be manufactured artificially. Look for evidence that the people behind a petition or boycott are real people and the effort is organic. Social media posts shouldn’t come from obscure users or bots. There’s a difference between a few snarky tweets and an actual public outcry.

21. THE WRITERS CREDENTIALS. Does the writer possess specialized knowledge in an area, either advanced education or experience covering a particular beat? This is especially important for opinion pieces. A list of qualifications in a writer’s bio should inform the audience as to their expertise. If they are not knowledgeable about the topic, they rely on other sources. An article without quotes or information from experts raises red flags, especially if creators present themselves as experts when they’re actually enthusiasts. 

22. ODD WORDING. AI-created text lacks the ability to write in a way that sounds natural—for now. When word choices or sentence structures are odd, it could be the product of a computer. The same is true with repeatedly using the same words and phrases or writing that lacks emotion. These are signs of machine-generated text.

OpenAI’s GPT-2 Detector is a tool that help to identify whether text is likely AI generated. While AI writing tools are not useful for reliable scientific texts without strong human intervention, it is likely predatory journals will exploit the quick production of scientific articles to generate low-quality content.

23. IMPROBABLE PRECISION. It’s a red flag when someone claims, "I drove from Chicago to Miami in 1.5847 hours." That kind of precision is unlikely.

More about spotting fake news

30 AI & Data Science Articles form July

Using data from high-resolution satellite images. USC researchers say have developed an AI model that they say accurately predicts where a wildfire will go next

A Look Under the Hood of Scopus AI: Elsevier’s search tool for scholarly testing

Bayesian Thinking in Modern Data Science

When AI makes a fatal mistake, who’s to blame? Air Force Secretary weighs morality and reality

An AutoML Framework Designed to Synthesize End-to-End Multimodal Machine Learning ML Pipelines Efficiently 

Tracking AI-Related Activities in the Private Sector 

A collection of tutorials focused on visualizing data

Could TTT models replace transformers to become the next frontier in generative AI? 

US spy satellite agency rethinks its relationship with Space Force

AI definitions: Algorithms

Direct, specific instructions for computers created by a human through coding that tells the computer how to perform a task. This set of rules has a finite number of steps that instruct the computer how to perform a task. More specifically, it is code that follows the algorithmic logic of “if”, “then”, and “else.”   

Do LLMs need a strong model of the world to get past its limitations? This author argues that isn’t the issue.

How to assess a general-purpose AI model’s reliability before it’s deployed

Big Data Analytics in 2025: Top Trends to Watc

A Look at the technology & hardware that make spy satellites work 

The US military will award a $3 billion contract for AI-driven geospatial intelligence this fall

“For every ML model that you consider deploying, make sure that your data scientists provide you with a full view of its potential business value”

Tools every data scientist should know: A practical guide

AI Definitions — Agents

Unlike AI prompts requiring user conversations, AI agents work in the background. Users provide a goal (from researching competitors to buying a car) and the agent acts independently, generating task list and starting to work. The industry is working toward a goal of agents operating independently. More AI definitions here:  

“Trump’s allies are drafting a sweeping AI executive order that would launch a series of “Manhattan Projects” to develop military technology

AI model harnesses physics to autocorrect remote sensing data

10 GitHub Repositories to Master Data Science 

A Graph Reinforcement Learning-Based Handover Strategy for Low Earth Orbit Satellites under Power Grid Scenarios

AI definition: Digital Twin

This means replicating the physical in a virtual environment. The twin might be a copy of us or the objects around us, such as a video avatar of a person or a statistical model of a complex phenomenon...  

15 Graphs That Explain the State of AI in 2024

“What do we need from a probabilistic programming language to support Bayesian workflow?”

Certifications that can boost your data science career in 2024

The case for decentralized data scientists 

AI Definitions: Retrieval augmented generation (RAG)

This coding technique instructs the bot to cross-check its answer with what is published elsewhere, essentially helping the AI to self-fact-check. RAG lets companies “ground” AI models in their own data, ensuring that results come from documents within the company.    

Satellite Images Show Expansion of Suspected Chinese Spy Bases in Cuba

Artificial intelligence can speed-sort satellite photos

Helping nonexperts build advanced generative AI models

Why AI systems like ChatGPT are so good at some tasks and so bad at others

Researchers say they have created a machine learning algorithm that processes satellite data to more accurately & efficiently predict space weather conditions caused by solar activity

The capacity to be an “I”

Differentiation means the capacity of a family member to define his or her own life's goals and values apart from surrounding togetherness pressures, to say “I” when others are demanding “you” and “we.” It includes the capacity to maintain a (relatively) nonanxious presence in the midst of anxious systems, to take maximum responsibility for one's own destiny and emotional being. It can be measured somewhat by the breath of one's repertoire of responses when confronted with crisis. The concept should not be confused with autonomy or narcissism, however. Differentiation means the capacity to be an “I” while remaining connected.

Edwin Friedman, Generation to Generation

What Fake News is NOT

Some people will mislabel rumors, hoaxes, and real news stories they don’t like as “fake news.” Another area of confusion is stories that result from mistaken or bad journalism.  

Sometimes well-respected news organizations get it wrong: sources can lie, documents can be faked, and reporters can mishear quotes. Sometimes new information changes the basic understanding of what is known publicly. You wouldn’t call this fake news since the motivation for posting the original (but mistaken) information wasn’t to deceive. What can make the situation worse: is the financial pressure of shifting away from legacy media (like newspapers) into the digital world, leaving the news industry scrambling to figure out how to support quality journalism financially. 

Between the pressure to meet social media engagement quotas and competition with other publications, writers often don’t get the necessary time to craft thoughtful and nuanced stories—or possess the power to reject an assignment over concerns about amplification.

Inaccurate details, such as reporting that four people are dead in a plane crash instead of six, can result from an honest mistake. The wrong number might have been heard or written down.

During breaking news, information will quickly shift as bits trickle into news organizations. It takes time to get a clear overall picture of what’s happening. Sometimes law enforcement officials or public relations professionals get the story wrong and send inaccurate information. At those times, news organizations are simply repeating mistakes. This is most likely to happen when only one source of information available whenever a story breaks.

Legitimate news sources will report the truth—as best they know it at the time. But as new information comes in, the story can shift. Just like with scientific research, this meandering pathway is just part of the process of getting to the truth.

It’s worth noting that the approach of legacy news organizations (The Washington Post, CNN) differs from new media outlets (BuzzFeed News, Politico). Traditional outlets aim at objectivity or neutral-voice reporting, where the focus is on being balanced, keeping the journalist’s opinions out of reports. More recently launched news sites are likely to focus on immediacy and transparency over neutrality and update readers whenever more information is known. Each approach presents different weaknesses for reporters to overcome. Of course, commentators may reference news information but are not acting as neutral reporters. Opinion pieces are often confused with basic news reporting. Pay attention whether you are reading a news report, an editorial, a guest blogger, a review, a disguised ad, or a comment.

The bottom line: be skeptical and bring a critical mind with you to everything you read. Keep in mind that “fake news” can be about something else besides the truth. As University of Southern California media scholar Mike Ananny has said, it is often “a struggle between [how] different people envision what kind of world that they want.”

Here are some tips for determining if a story is likely reliable. An organization does not need to tick off all these qualifiers in order to be considered authentic and accurate, but the more you see red flags pop up, the more a healthy skepticism is in order.

You’ll notice these are “tips” and not a checklist. Checklists can oversimply the nuances of discernment into black-and-white boxes.

More about spotting fake news