Identifying Election Day Misinformation Trends

Excellent (and terrifying) new analysis by Accountable Tech on election preparedness of social media platforms, including a scorecard measuring to what extent their policies meet recommendations made in the Democracy By Design roadmap – actionable, high-impact, and content-agnostic steps to protect the integrity of elections. Download here: https://lnkd.in/eNWHnHRE Thanks Nicole Gill and team for your hard work and insights. TOPLINES: 1. Out of a possible 100% match to the Democracy By Design recommendations, no platform scores above 62%. Nextdoor performs the worst, with a 17% preparedness score 2. There are insufficient guardrails to stop the spread of manipulated content depicting public figures, like deepfakes: Just 20% of platforms – TikTok andSnapchat – have policies on the books that would prohibit deceptively manipulated media of public figures. 3. Platform features enable AI-generated political ads to be micro-targeted to voters: Nearly every social media platform which allows political advertising does not explicitly prohibit AI-generated ads from being micro-targeted to voters. 4. There is a lack of transparency on performance and engagement related to election-related posts: No platform provides transparent access to data related to the highest-performing and highest-engagement election-related posts, advertisements, accounts, URLs, and groups. That means that voters, independent researchers, and election officials are left in the dark about how election-related information spreads across platforms.   5. Insufficient “friction” to stop the spread of misleading election information: A majority of platforms do not have policies in place to put posts that contain misleading or unverified election information behind click-through warning labels that include clear context and fact. Without these labels, election misinformation is able to be spread more quickly and magnify threats. 6. There's a lack of transparency, including an opacity around policy enforcement and safety teams: Some platforms, like Meta, which have previously come under intense scrutiny for their role in amplifying the spread of electoral disinformation narratives, have numerous policies, but it’s impossible to know how they are being enforced. Platforms have wide latitude when it comes to enforcement, and there is reason for skepticism that they meaningfully follow through. This is made more concerning because of industry-wide layoffs and cuts to election integrity safety teams – including the complete dismantling of X’s election integrity team.

...ding, ding, ding - another #disinformation update. And again: German #elections (#Bundestagswahl25). ISD (Institute for Strategic Dialogue) identified a coordinated disinformation network using AI and media impersonation to target German elections. "A #bot network of more than 6,000 accounts amplifies videos, often tagging media outlets or fact-checkers in their posts. The volume of new videos increased dramatically in the second half of January 2025. However, this increased output has not resulted in more organic engagement, with bots responsible for nearly all shares. Despite focusing on the German election, none of the videos or posts are in German. This suggests the network intends to overwhelm fact-checkers and mislead international observers rather than influence Germans themselves." Here is the link to the report: https://lnkd.in/e9iUx_gd Here some Key Findings - A primary network of at least 48 X accounts. This network became active in November 2024, initially focusing on allegations of rising antisemitism in Germany. Since 10 January 2025, it turned its attention to the elections. - A secondary network of more than 6,000 accounts is simultaneously reposting the content, ensuring that the original network’s videos are shared thousands of times within minutes. - Recent videos feature disinformation about election-related terror threats and false claims about German politicians, particularly allegations of corruption and paedophilia - Main targets of the campaign are Friedrich Merz (CDU Deutschlands); Janine Wissler (Die Linke); and Armin Laschet, the former CDU chancellor candidate. - Videos shared by the network carry branding from legitimate media organisations (e.g. Deutsche Welle (DW), BBC and Sky News). They also imitate government agencies and academic institutions. - The operation has impersonated at least 20 organisations since the beginning of 2025, at times using AI to manipulate the audio of real videos or adding captions featuring false claims. - The network’s tactics – impersonating legitimate organisations in videos, using AI, posting QR codes, and tagging journalists and fact checkers – are commonly used by a Russia-aligned information operation known as ‘#OperationOverload,’ (also "#Matroyshka"). - The campaign has limited impact, with the majority of its engagement coming from this secondary bot amplification network. However, the 48 accounts ISD initially identified have collectively received 2.5 million views, with engagement tripling in January. - The network is sharing content in several languages, including English, Spanish, and Arabic, but not in German. - Thus, the aim of this campaign is to undermine trust in German elections among international audiences.

https://lnkd.in/e7WJxCEb *Note:"PREBUNKING," not debunking, activities can generate positive effect. "Despite its January 13 election being assailed by a deluge of online disinformation — particularly, false voter fraud claims and dire warnings of future war from bad actors in China — new research and independent journalistic accounts reveal that local media, election authorities, and fact checkers in Taiwan were largely successful in repelling assaults, with techniques such as 'prebunking,' smart communications regulation, and a deliberate focus on media trust. After the vote, Lai Ching-te — head of the pro-sovereignty Democratic Progressive Party, which is opposed by Communist China’s autocratic government — was elected president. However, Taiwan’s election revealed disinformation trends for journalists and fact checkers in other countries to flag. These patterns included the use of generative AI in deepfakes, propaganda amplification by popular YouTube influencers, and foreign information operation narratives designed to undermine trust in democracy itself, rather than to promote individual candidates."

Last week, I joined Jay Strubberg on Morning Rush to address a pressing issue: the growing threat of deepfake videos in elections. This discussion was spurred after Elon Musk shared a digitally altered AI video of Vice President Kamala Harris, which has over 100 million views without any context indicating it was fake.   The problem of manipulated AI videos used to spread misinformation about political candidates is not confined to the U.S.; it's a global concern that will impact elections worldwide unless we establish an international coalition that will work to ensure the integrity and authenticity of online content.   Given that we won't see a policy solution by the November election, it's crucial for campaigns and experts to educate the public on how to identify manipulated videos. Here are some tips for voters to consider:   Verify the Source: Always check the origin of the video and whether it comes from a credible source. Cross-Check Information: Compare the message with what you know about the candidate to spot inconsistencies. Question Suspicious Content: If something seems off or doesn't add up, take a step back and verify the information before accepting it as true.   By staying informed and vigilant, we can better protect the democratic process from the threats posed by deepfake technology.

A recent study by researchers at The Alan Turing Institute and University of Oxford delves into how effectively #LargeLanguageModels (#LLMs) can be used for creating #electiondisinformation and how well #humans can detect such content (https://lnkd.in/dT8zzAv2). While previous studies examined AI-generated #politicalmisinformation, this research is unique in exploring both AI’s willingness to generate disinformation (AI capability layer) and human success in identifying it (human-interaction layer). Using 13 AI models, from older to the latest versions (such as T5 from 2019 to Llama 3 from 2024), and both open-source and closed-source models, the study tested each model’s responses to prompts for creating #fakenews articles, fake #socialmedia accounts, #posts, and replies. Some prompts involved specific disinformation scenarios with details like #PoliticalParties and alleged offences. Individual human detection abilities varied, but demographic factors (e.g., education) didn’t impact #Accuracy. Surprisingly, some AIs created news articles that seemed more authentic than actual human-written content, likely due to their polished, structured style. The study reveals that human reviewers, especially older and right-wing-identifying individuals, were more prone to misidentify AI-generated content, particularly if the content had a left-wing #bias. Researchers warned that biases in AI responses could fuel perceptions of political censorship, as seen in recent U.S. accusations of #TechCompanies censoring conservative content. While recent models show some progress in refusing disinformation requests, the gap between AI’s persuasive abilities and the effectiveness of these guardrails is widening. The study concludes that achieving complete safety against AI-driven misinformation is unlikely through AI improvements alone. Instead, a broader #AILiteracy among users and institutions is needed to address AI’s growing influence on information operations (https://lnkd.in/dfMmmHFw). With developers racing to make AIs ever more convincing, the challenge lies in educating the public to discern the real from the machine-generated in an increasingly complex information landscape, because some humans also have no ethical constraints in spreading false and defamatory news.

🔭Next time you hear someone say “it’s just coms”, pull out the striking visuals provided by the European External Action Service (EEAS) in their last report on information threats / #FIMI ! 🗞️ This 3️⃣rd yearly report focuses on the Architecture of interference, mapping the tactics and visualizing the intent of manipulative threat actors. It sheds light on the comprehensive & multi-layered digital architecture used by authoritarian regimes, esp. Russia 🇷🇺 and China 🇨🇳 to conduct their operations. 🚦Its key message : 🔹Overt and covert channels/media outlets work together to produce operations 🔹threat actors think of this as one unique field 🔹 siloed approaches don’t capture this 🔹making the link btw online channels and threat actors is the basis for attribution & holding them accountable 🔹policy makers must be aware of how official & attributed channels interact with extensive covert network of state-linked channels hidden from the public eye 📊Among the findings : 🔹Different modus operandi of Russian and Chinese FIMI operations, but mutual amplification to reinforce anti-Western messaging 🔹Half of the attacks targeted Ukraine. 🔹France, Germany, Moldova and Sub-Saharan Africa, notably the Sahel also heavily targeted 🔹Elections a key target : 42 Russian attempts recorded during 🇪🇺 elections 🔹institutions attacked too : #EU, #NATO 🔹independent media outlets and individual FIMI defenders among most attacked 📈 Trends 🔺Most activity on social media 🔺X = 88% of detected activity 🔺Key tactics= bot networks, impersonation and creation of inauthentic news websites, Advances in generative Artificial Intelligence use to increase scale 🔎 Methodology : 🔹505 FIMI incidents analysed in 2024, providing an overview of key FIMI trends 🔹38,000 channels 🔹Focus of vast online infrastructure used by Russia and China 🔹Case studies : Russian campaigns in Moldova and Africa, an operation originating from China Conclusion ▪️#FIMI networks adapt their strategies to geopolitical shifts and tailor their tactics to local contexts. ▪️Good tools here to empower citizens in understanding how the integrity of their info space is affected 👏🏼 #EEAS colleagues ! Aude Maio-Coliche Filip Grzegorzewski An inspiration for Sabrina Spieleder & team to keep up the good work 🎯