How Social Media and Artificial Intelligence Are Transforming Information, Trust, and Our Decisions
By Judit Vitores-Sypher
For generations, a photograph was considered evidence of an event. An audio recording seemed to confirm a statement, and a video allowed us to witness what had happened with our own eyes. Today, those certainties are changing.
Artificial intelligence can fabricate images of events that never occurred, reproduce human voices, and generate videos in which real people appear to say or do things they never said or did. Meanwhile, social media platforms can distribute such material to enormous audiences within minutes.
We live with a paradox: never before have we had so much access to information, yet distinguishing truth from manipulation requires increasingly greater effort. Are we truly informed, or are we forming opinions based on messages delivered to us through recommendation systems we barely understand?
From Social Media to Algorithms
Facebook, Instagram, TikTok, YouTube, and other platforms have transformed everyday communication. What began as personal exchanges now also functions as a channel for news, public debate, advertising, and opinions.
This has expanded opportunities for expression, but it has also changed how we receive information.
According to the Reuters Institute's Digital News Report 2026, across an average of 48 markets, 54% of respondents used social media and video platforms to access news during the previous week, while 51% turned to news websites and apps. Meanwhile, 62% expressed concern about false information online.
These percentages measure different indicators and do not represent mutually exclusive categories: the same person may obtain news through different channels while also expressing concern about disinformation. [1]
Algorithms select posts based on multiple signals of interest and engagement. As a result, two people may encounter very different narratives about the same event.
A message that provokes strong emotions and is widely shared is not necessarily true. Popularity and credibility are two different things.
When Fiction Appears to Be Reality
Propaganda, rumors, and altered images existed long before artificial intelligence. What is new is the speed, relatively low cost, and realism with which falsifications can now be produced.
Deepfakes are synthetic or manipulated audiovisual materials capable of convincingly imitating voices, faces, and actions.
Not everything created with artificial intelligence is deceptive. This technology has legitimate applications in education, accessibility, and journalism.
The problem arises when fabricated content is falsely presented as a real event.
Deception can also occur without AI: old photographs published with misleading dates, edited videos, and statements taken out of context.
Documented Case: The Fake Pentagon Explosion
On May 22, 2023, an image circulated online supposedly showing an explosion near the Pentagon in the United States.
The incident never occurred.
AFP Fact Check published a verification the following day, reporting denials from the Department of Defense and Arlington firefighters. It also identified signs suggesting that the image had been artificially generated. [2]
As the rumor spread, financial markets experienced brief fluctuations.
The timing alone does not prove that every market movement resulted from that single publication. Nevertheless, the incident illustrates how a misleading image can trigger reactions before it is debunked.
Documented Case: A Fake Video Call and a Multimillion-Dollar Scam
In 2024, reports emerged that an employee of the international engineering firm Arup had been deceived in Hong Kong during a video conference featuring digital impersonations of executives and colleagues.
The employee made transfers totaling approximately HK$200 million, equivalent at the time to roughly US$25.6 million.
The company confirmed the fraud. [3]
Recognizing a face and voice on a screen is no longer sufficient to guarantee someone's identity.
Payment instructions and requests for sensitive information require independent verification.
Other Cases That Raised the Alarm
In January 2024, voters in New Hampshire received automated telephone calls featuring an AI-generated voice designed to sound like Joe Biden.
The message sought to discourage participation in the presidential primary election.
The recording did not originate from the then-president and was denounced as an attempt at election interference. [4]
Disinformation and Digital Misogyny: When Appearance Becomes a Weapon
Women who participate in public life may face attacks that shift attention away from their decisions and toward their bodies, clothing, or physical appearance.
Gender-based disinformation, studied by researchers and organizations including UNESCO, combines misleading claims with sexist stereotypes to undermine women's credibility and public voices. [6]
A recent example of this dynamic involves posts from October 2026 and earlier comments comparing Kelly Loeffler, administrator of the U.S. Small Business Administration (SBA), and Susan Dell, philanthropist and wife of businessman Michael Dell, to supposed extraterrestrial beings.
Mocking comments about Barbera Thornhill also circulated.
News coverage and examples of public posts document physical comparisons and certain conspiracy claims. However, they do not, by themselves, establish that the photographs were digitally altered or that a coordinated campaign existed.
Susan Dell is not a member of the federal government. [7, 13–21]
That distinction matters.
An authentic photograph can be used to promote an unfounded interpretation.
Criticizing a public official's policies is part of democratic debate. Attributing a nonhuman identity to that person without evidence, or reducing her to ridicule based on physical appearance, provides no meaningful information about her performance in office.
The journalistic question is what is being shared, who originated it, and how an insinuation becomes an apparent certainty.
Another example illustrates the consequences of digital manipulation for personal privacy.
In January 2024, fabricated sexually explicit images of singer Taylor Swift, generated using artificial intelligence, circulated online without her consent.
The incident exposed the harm such content can cause to privacy and reputation—harm that can also affect people outside the public spotlight. [5]
But manipulation through artificial intelligence is not limited to attacks on someone's image or reputation.
It can also exploit deeply human emotions, including the love people feel for their pets and the anguish caused by their disappearance.
Documented Case: Lost-Pet Scams
Digital manipulation can also infiltrate virtual neighborhood communities.
Posts about missing pets, shared through social media and neighborhood applications, provide scammers with information and photographs they can use to fabricate false evidence that an animal has been found.
Their objective is to obtain money from distressed owners desperately hoping to recover a beloved companion.
In April 2026, San Francisco Bay Area media reported the case of Alisa Harrell, an Oakland resident who received what appeared to be a recent photograph of her missing dog, Dash, accompanied by a demand for payment.
Harrell consulted neighbors on Nextdoor, who identified signs suggesting the image might have been generated using artificial intelligence.
She did not pay.
An important distinction should be made: in this case, the neighborhood platform also helped expose the suspected fraud. [13]
On September 30, 2026, police in Norwalk, Connecticut, warned about a similar scheme.
Scammers were reportedly taking photographs of missing pets posted on social media, generating synthetic images, and then demanding money or gift cards from owners in exchange for supposedly returning their animals. [14]
The vulnerability being exploited is not technological but profoundly human: the hope of finding a missing pet.
Before sending money, owners should request independent, verifiable proof, confirm details that only someone genuinely in possession of the animal would know, and arrange any meeting in a safe location.
A convincing photograph, by itself, is no longer sufficient proof.
Although these cases differ, they demonstrate how convincing content can exploit trust and emotions—sometimes to discredit individuals and sometimes to obtain money.
Why We Believe and Share False News
Emotions influence how we evaluate what we read.
Confirmation bias makes us more likely to pay attention to messages that align with our existing beliefs. Familiarity with the person sharing information may also lead us to confuse personal trust with journalistic verification.
On platforms filled with constant stimulation, we often react before examining the context.
A story repeated many times may begin to seem more credible without any additional supporting evidence.
Sharing something with good intentions does not eliminate the potential harm caused by inaccurate information.
Decisions Under the Influence
False information can affect medical consultations, investments, purchases, relationships, and opinions about public events.
Misleading health advice can delay treatment. A cloned voice can facilitate fraud. A fabricated accusation can damage someone's reputation.
The proliferation of falsified content also creates what researchers call the "liar's dividend": the possibility of discrediting an authentic recording by claiming, without evidence, that it was created using artificial intelligence.
Deception can therefore lead us both to believe events that never occurred and to distrust genuine evidence.
Learning How to Verify Information
There is no single, infallible detection tool.
To verify images, it is useful to trace earlier publications through reverse-image searches.
For videos, we should locate the complete original version and verify the date, location, and context.
For sensitive audio recordings, confirming the speaker's identity through an independent channel is advisable.
Digital provenance credentials can provide information about a file's origin when available. Their absence, however, does not mean the content is false.
Primary sources and professional fact-checking organizations remain essential.
FIVE STEPS TO DEBUNK FALSE INFORMATION
1. Investigate the source. Identify who published the information and whether the author has a verifiable identity and track record.
2. Recover the context. Look for the date, location, and original publication. Even an authentic photograph can accompany a fabricated story.
3. Cross-check accounts. Seek corroboration from independent sources and professional fact-checkers.
4. Verify visual content. Use tools such as Google Lens or TinEye and examine the original material. An unusual appearance does not, by itself, prove manipulation.
5. Pause before sharing. If a message triggers fear, outrage, or urgency, examine the evidence. If the evidence is insufficient, do not present the claim as fact.
Doubt opens the door to investigation; it does not, by itself, constitute proof that something is false.
Digital Education and Shared Responsibility
Teaching people how to operate devices is not enough.
We need media literacy, critical thinking, and an understanding of how recommendation systems work.
News organizations must investigate and correct errors. Digital platforms need greater transparency. Every user can contribute by avoiding the spread of unverified claims.
Truth Should Not Depend on an Algorithm
As a journalist, I believe one of our greatest challenges is not merely the advancement of artificial intelligence, but how quickly we have grown accustomed to consuming information without questioning it.
We observe, react, and share.
But how often do we stop to verify who published something, where it happened, and what evidence actually exists?
Technology can help us learn, investigate, and understand.
It should not replace our judgment.
Changing our minds when better evidence emerges is not a weakness. It is an exercise in critical thinking.
The real danger is not simply that technology can fabricate convincing lies. It begins when we stop asking ourselves what is true.
Sources and Editorial References
[1] Reuters Institute. Digital News Report 2026.
https://reutersinstitute.politics.ox.ac.uk/digital-news-report/2026/dnr-executive-summary
[2] AFP Fact Check. May 23, 2023. Verification of the fabricated Pentagon explosion image.
https://factual.afp.com/doc.afp.com.33FY67K
[3] South China Morning Post. May 17, 2024. Arup confirms the Hong Kong deepfake fraud.
https://www.scmp.com/news/hong-kong/law-and-crime/article/3263151/uk-multinational-arup-confirmed-victim-hk200-million-deepfake-scam-used-digital-version-cfo-dupe
[4] Associated Press. January 2024. AI-generated Biden robocalls.
https://apnews.com/article/f3469ceb6dd613079092287994663db5
[5] Associated Press. January 26, 2024. Deepfake explicit images of Taylor Swift spread on social media. Her fans are fighting back.
https://apnews.com/article/d5eb3f98084bcbb670a185f7aeec78b1
[6] UNESCO. May 2024. Women as Targets of Gender-Based Misinformation.
https://www.unesco.org/en/articles/women-targets-gender-based-misinformation
[7] Rasmus Senator, Newsner. October 16, 2026. After Susan Dell's Viral Transformation, Video of Top Official Fuels Speculation Over Women in Trump's Circle. Documents online comments, not digital alteration.
https://en.newsner.com/news/trump-official-kelly-loeffler-video/
[8] Hong Kong Government. June 26, 2024. Official response concerning deepfake scams.
https://www.info.gov.hk/gia/general/202406/26/P2024062600192.htm
[9] Associated Press / PBS NewsHour. May 23, 2024. Political Consultant Behind AI-Generated Biden Robocalls Faces $6 Million Fine and Criminal Charges.
https://www.pbs.org/newshour/politics/political-consultant-behind-ai-generated-biden-robocalls-faces-6-million-fine-and-criminal-charges
[10] Federal Communications Commission. 2024. Official proceedings concerning fraudulent Biden robocalls.
https://www.govinfo.gov/content/pkg/GPO-FCC-Rcd-V39No12/pdf/GPO-FCC-Rcd-V39No12.pdf
[11] Wilson Center. January 25, 2021. Malign Creativity: How Gender, Sex, and Lies Are Weaponized Against Women Online.
https://www.wilsoncenter.org/publication/malign-creativity-how-gender-sex-and-lies-are-weaponized-against-women-online
[12] Chesney, Robert, and Danielle Citron. 2019. Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security. California Law Review, 107(6), 1753–1820.
[13] KTVU FOX 2. April 21, 2026. Reporting on Alisa Harrell and her missing dog, Dash.
https://www.ktvu.com/news/heartbroken-bay-area-pet-owner-searching-lost-dog-says-scammers-used-ai-try-extort-money
[14] Patch. October 1, 2026. Norwalk Police warning concerning AI-generated lost-pet scams.
https://patch.com/connecticut/norwalk/norwalk-police-warn-residents-ai-generated-lost-pet-scam-pd
Editorial note: The examples involving Kelly Loeffler, Susan Dell, and Barbera Thornhill concern documented online comments and conspiracy-related comparisons. They do not establish that the photographs were digitally altered or that the posts were part of a coordinated campaign.
© 2026 Judit Vitores-Sypher.
