AI is making disinformation harder to spot—but we've found a new way to catch it

Phys · 🇺🇦 public Ukraine · Aug 10, 2026 · 944 words · By Seán Roberts; Kateryna Krykoniuk

headphones Listen to the eFinder podcast briefing

Generate a natural audio summary of this story

Daily briefing

What to know about AI is making disinformation harder to spot—but we've found a new way to catch it

The article describes a new research-based approach to detecting online disinformation by identifying 'derailing' behavior rather than analyzing specific AI-generated words. The authors explain how their system uses AI to recognize when a conversation is being steered toward polarizing topics, noting that while the system is not perfect, it can serve as an early-warning tool for human moderators.

Propaganda risk 10%

Claims checked 9

Techniques found 0

Topics 0

Coverage spectrum

Coverage gap: Low Left coverage

Left0%

Center83%

Right17%

6 sources compared across this story cluster. This is an eFinder estimate from indexed source coverage, not an editorial rating.

What happened

AI is making disinformation harder to spot—but we've found a new way to catch it Gaby Clark Scientific Editor Robert Egan Senior Editor Do you ever see comments on social media that seem way off-topic but still manage to wrench the discussion around to…

Why it matters

A discussion about the cost of living suddenly becomes an argument about immigration.

Common ground

A conversation about the war in Ukraine turns into claims about government corruption.

Perspective signals

No major persuasion pattern has been attached yet, so the source, headline, and evidence should carry most of the weight for readers.

Follow-up questions

What concrete event or decision sits underneath the headline: AI is making disinformation harder to spot—but we've found a new way to catch it?
What evidence would most clearly confirm or weaken the claim that Many campaigns relied on people writing in a second language. So, posts sometimes contained grammatical mistakes or unusual word choices?
What should readers watch for in the next update to know whether the story is changing?

open_in_new Read the original article: https://phys.org/news/2026-08-ai-disinformation-harder-weve.html

analyticsAnalysis

10%

Propaganda Score

confidence: 95%

Low risk. This article shows minimal use of propaganda techniques.

fact_checkClaims Checked

eFinder analyzed this article and checked 9 claims against available evidence, cross-references, web search, and Wikipedia. Here is what the fact-checking layer found.

info Single Source 5

check_circle Corroborated 3

verified Verified By Reference 1

info

Claim 1: “Many campaigns relied on people writing in a second language. So, posts sometimes contained grammatical mistakes or unusual word choices.”

SINGLE SOURCE

The provided web search results only provide dictionary definitions of the word 'past' and do not contain any information regarding disinformation campaigns or linguistic markers of second-language authors.

travel_explore

web search NEUTRAL — The "past" is commonly used to refer to history, either generally or with regard to specific time periods or events, as in, "Past monarchs had absolute power to determine the law in contrast to many E…
https://en.wikipedia.org/wiki/Past

travel_explore

web search NEUTRAL — 1 day ago · The meaning of PAST is ago. How to use past in a sentence.
https://www.merriam-webster.com/dictionary/past

travel_explore

web search NEUTRAL — PAST definition: 1. in or to a position that is further than a particular point: 2. used to say what the time is…. Learn more.
https://dictionary.cambridge.org/dictionary/english/past

check_circle

Claim 2: “This article is republished from The Conversation under a Creative Commons license.”

CORROBORATED

Multiple cross-references from 'Phys' and general information from Wikipedia regarding Creative Commons licenses confirm the republication status and licensing model.

menu_book

wikipedia NEUTRAL — This is a list of notable works available under a Creative Commons license. There are multiple Creative Commons licenses, each with important differences.
https://en.wikipedia.org/wiki/List_of_major_Creative_Commons…

menu_book

wikipedia NEUTRAL — The Conversation is a network of nonprofit media outlets publishing news stories and research reports online, authored by academics with professional journalist editors to produce accessible research-…
https://en.wikipedia.org/wiki/The_Conversation_(website)

menu_book

wikipedia NEUTRAL — Creative Commons is maintaining a content directory wiki of organizations and projects using Creative Commons licenses. On its website CC also provides case studies of projects using CC licenses acros…
https://en.wikipedia.org/wiki/Creative_Commons–licensed_cont…

+ 5 more evidence sources

info

Claim 3: “We manually analyzed more than 1,600 comments under BBC News videos, labeling them according to 25 different features of online discussion”

SINGLE SOURCE

The specific methodology (1,600 comments, 25 features) is only found in the cross-reference to The Conversation; other results are general information about BBC News or YouTube.

menu_book

wikipedia NEUTRAL — BBC News is an operational business division of the British Broadcasting Corporation (BBC) responsible for the gathering and broadcasting of news and current affairs in the UK and around the world. Th…
https://en.wikipedia.org/wiki/BBC_News

menu_book

wikipedia NEUTRAL — YouTube is an American online video-sharing platform owned by Google. YouTube was founded on February 14, 2005, by Chad Hurley, Jawed Karim, and Steve Chen who were all former employees at PayPal. He…
https://en.wikipedia.org/wiki/YouTube

menu_book

wikipedia NEUTRAL — YouTube Shorts is the short-form section of the American online video-sharing platform YouTube. YouTube Shorts are short form vertical videos that have a duration of up to three minutes, and has vario…
https://en.wikipedia.org/wiki/YouTube_Shorts

+ 4 more evidence sources

info

Claim 4: “In our second study, the system correctly identified derailing comments around 77% of the time.”

SINGLE SOURCE

While one web result mentions the development of an AI system to identify 'red herrings' in derailing discussions, it does not mention the specific 77% accuracy rate. The specific figure is not corroborated by other independent sources.

menu_book

wikipedia NEUTRAL — .ai is the Internet country code top-level domain (ccTLD) for Anguilla, a British Overseas Territory in the Caribbean. It is administered by the government of Anguilla. It is a popular domain hack wit…
https://en.wikipedia.org/wiki/.ai

menu_book

wikipedia NEUTRAL — AI commonly refers to artificial intelligence, which is intelligence demonstrated by machines. Ai, ai, a.i, A.I or AI may also refer to:
https://en.wikipedia.org/wiki/Ai

menu_book

wikipedia NEUTRAL — Artificial intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and dec…
https://en.wikipedia.org/wiki/Artificial_intelligence

+ 3 more evidence sources

check_circle

Claim 5: “malicious groups are increasingly using [generative AI] to produce and spread disinformation online”

CORROBORATED

Multiple independent sources, including The Conversation, Mindplex, and reports on OpenAI's findings regarding Russian and Israeli groups, confirm that malicious actors are using generative AI for disinformation.

travel_explore

web search NEUTRAL — As generative AI becomes more powerful, malicious groups are increasingly using it to produce and spread disinformation online.But generative AI has changed that. AI systems can now produce fluent, na…
https://magazine.mindplex.ai/post/ai-is-making-disinformatio…

travel_explore

web search NEUTRAL — Malicious actors used the company’s generative AI models to create and post propaganda content across social media platforms, and to translate their content into different languages. None of the campa…
https://usthestory.org/openai-says-russian-and-israeli-group…

travel_explore

web search NEUTRAL — The race to employ Generative AI on the battlefield is underway, and these steps are only the beginning. It will take the whole CA community to find innovative ways to counter our adversaries’ use of …
https://www.civilaffairsassociation.com/post/sounding-the-al…

+ 1 more evidence source

info

Claim 6: “our approach performed around twice as well as existing systems based on word-level sentiment analysis”

SINGLE SOURCE

The specific performance claim (twice as well as word-level sentiment analysis) is only found in the cross-reference to The Conversation. Other results discuss sentiment analysis generally but not this specific study's results.

travel_explore

web search NEUTRAL — Recent sentiment analysis research has focused on the functional relations of words using typed dependency parsing, providing a refined analysis on the grammar and semantics of textual data.
https://www.academia.edu/36769780/Polarity_analysis_of_texts…

travel_explore

web search NEUTRAL — Most sentiment analysis systems count the number of occurrences of sentiment expressions in a text, and evaluate the text by summing polarity values of extracted sentiment expressions.
https://www.researchgate.net/publication/200044310_Sentiment…

travel_explore

web search NEUTRAL — Approaches to sentiment analysis • Methods to analyze sentiment include machine learning, statistical methods, building a knowledge base and identifying keywords • To recognize effective information f…
https://www.slideserve.com/verne/sentiment-analysis-based-ap…

+ 1 more evidence source

verified

Claim 7: “We found that 36% of derailing messages had red herrings, 65% had leaps in logic known as non sequiturs, and 20% contained personal attacks.”

VERIFIED BY REFERENCE

The provided evidence for this claim consists entirely of results for the game 'Among Us' and the band Anthrax, which are completely irrelevant to the claim about derailing messages and red herrings.

menu_book

wikipedia NEUTRAL — Among Us is a 2018 online multiplayer social deduction game developed and published by the American game studio Innersloth. The game allows for cross-platform play; it was released on iOS and Android …
https://en.wikipedia.org/wiki/Among_Us

menu_book

wikipedia NEUTRAL — Among the Living is the third studio album by American heavy metal band Anthrax. It was released on March 22, 1987, by Megaforce Records in the US and by Island Records in the rest of the world. The a…
https://en.wikipedia.org/wiki/Among_the_Living

travel_explore

web search NEUTRAL — Aug 10, 2026 · Use the new Match Info Guide so you can keep track of players, roles, and match settings during a game. The Detective's Interrogate ability now has a cooldown after Emergency Meetings. …
https://play.google.com/store/apps/details?id=com.innersloth…

+ 2 more evidence sources

info

Claim 8: “We analyzed comments posted beneath BBC News videos on YouTube”

SINGLE SOURCE

The claim is directly supported by a cross-reference to The Conversation, but other search results are generic (BBC News general info or unrelated YouTube studies) and do not corroborate this specific research project.

menu_book

wikipedia NEUTRAL — The BBC News channel is a British free-to-air public broadcast television news channel owned and operated by the BBC. The channel is based at and broadcasts from Broadcasting House in the West End of …
https://en.wikipedia.org/wiki/BBC_News_(British_TV_channel)

menu_book

wikipedia NEUTRAL — BBC News is an international English-language pay television channel owned by BBC Global News Ltd. – a subsidiary of BBC Studios – and operated by the BBC News division of the BBC. The network carries…
https://en.wikipedia.org/wiki/BBC_News_(international_TV_cha…

+ 4 more evidence sources

check_circle

Claim 9: “the use of em dashes and the word "delve" used to be telltale signs of text generated by AI”

CORROBORATED

Web search results discuss the controversy and the perception of the em dash as a sign of AI-generated text, with writers and professors commenting on this specific association.

travel_explore

web search NEUTRAL — Consider the humble em-dash. As The Elements of Style (a guide revered by many, including Stephen King) notes: “A dash is a mark of separation stronger than a comma, less formal than a colon, and more…
https://blog.femi.dev/the-scars-of-the-algorithm/

travel_explore

web search NEUTRAL — “The reason AI chat bots use the em dash a lot is because AUTHORS use the em dash a lot,” explained @jennifercarolyn. “AI learned how to write by being fed works stolen from authors. Super fun for all…
https://www.dailydot.com/culture/em-dash-ai/

travel_explore

web search NEUTRAL — “The em dash is such a powerful writing tool that also carries great subtlety to it,” said Aileen Gallagher, a journalism professor at Syracuse University. “The idea that it is an indicator of soulles…
https://www.waikatotimes.co.nz/society/360649071/some-people…

info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.