#FactCheck- AI-Generated Image Falsely Shows SRH Team Seeking Blessings

As AI language models become more powerful, they are also becoming more prone to errors. One increasingly prominent issue is AI hallucinations, instances where models generate outputs that are factually incorrect, nonsensical, or entirely fabricated, yet present them with complete confidence. Recently, ChatGPT released two new models—o3 and o4-mini, which differ from earlier versions as they focus more on step-by-step reasoning rather than simple text prediction. With the growing reliance on chatbots and generative models for everything from news summaries to legal advice, this phenomenon poses a serious threat to public trust, information accuracy, and decision-making.

What Are AI Hallucinations?

AI hallucinations occur when a model invents facts, misattributes quotes, or cites nonexistent sources. This is not a bug but a side effect of how Large Language Models (LLMs) work, and it is only the probability that can be reduced, not their occurrence altogether. Trained on vast internet data, these models predict what word is likely to come next in a sequence. They have no true understanding of the world or facts, they simulate reasoning based on statistical patterns in text. What is alarming is that the newer and more advanced models are producing more hallucinations, not fewer. seemingly counterintuitive. This has been prevalent reasoning-based models, which generate answers step-by-step in a chain-of-thought style. While this can improve performance on complex tasks, it also opens more room for errors at each step, especially when no factual retrieval or grounding is involved.

As per reports shared on TechCrunch, it mentioned that when users asked AI models for short answers, hallucinations increased by up to 30%. And a study published in eWeek found that ChatGPT hallucinated in 40% of tests involving domain-specific queries, such as medical and legal questions. This was not, however, limited to this particular Large Language Model, but also similar ones like DeepSeek. Even more concerning are hallucinations in multimodal models like those used for deepfakes. Forbes reports that some of these models produce synthetic media that not only look real but are also capable of contributing to fabricated narratives, raising the stakes for the spread of misinformation during elections, crises, and other instances.

It is also notable that AI models are continually improving with each version, focusing on reducing hallucinations and enhancing accuracy. New features, such as providing source links and citations, are being implemented to increase transparency and reliability in responses.

The Misinformation Dilemma

The rise of AI-generated hallucinations exacerbates the already severe problem of online misinformation. Hallucinated content can quickly spread across social platforms, get scraped into training datasets, and re-emerge in new generations of models, creating a dangerous feedback loop. However, it helps that the developers are already aware of such instances and are actively charting out ways in which we can reduce the probability of this error. Some of them are:

1. Retrieval-Augmented Generation (RAG): Instead of relying purely on a model’s internal knowledge, RAG allows the model to “look up” information from external databases or trusted sources during the generation process. This can significantly reduce hallucination rates by anchoring responses in verifiable data.
2. Use of smaller, more specialised language models: Lightweight models fine-tuned on specific domains, such as medical records or legal texts. They tend to hallucinate less because their scope is limited and better curated.

Furthermore, transparency mechanisms such as source citation, model disclaimers, and user feedback loops can help mitigate the impact of hallucinations. For instance, when a model generates a response, linking back to its source allows users to verify the claims made.

Conclusion

AI hallucinations are an intrinsic part of how generative models function today, and such a side-effect would continue to occur until foundational changes are made in how models are trained and deployed. For the time being, developers, companies, and users must approach AI-generated content with caution. LLMs are, fundamentally, word predictors, brilliant but fallible. Recognising their limitations is the first step in navigating the misinformation dilemma they pose.

References

• https://www.eweek.com/news/ai-hallucinations-increase/
• https://www.resilience.org/stories/2025-05-11/better-ai-has-more-hallucinations/ 
• https://www.ekathimerini.com/nytimes/1269076/ai-is-getting-more-powerful-but-its-hallucinations-are-getting-worse/ 
• https://techcrunch.com/2025/05/08/asking-chatbots-for-short-answers-can-increase-hallucinations-study-finds/ 
• https://en.as.com/latest_news/is-chatgpt-having-robot-dreams-ai-is-hallucinating-and-producing-incorrect-information-and-experts-dont-know-why-n/
• https://www.newscientist.com/article/2479545-ai-hallucinations-are-getting-worse-and-theyre-here-to-stay/
• https://www.forbes.com/sites/conormurray/2025/05/06/why-ai-hallucinations-are-worse-than-ever/ 
• https://towardsdatascience.com/how-i-deal-with-hallucinations-at-an-ai-startup-9fc4121295cc/ 
• https://www.informationweek.com/machine-learning-ai/getting-a-handle-on-ai-hallucinations 

‍

Executive Summary:

Our Team recently came across a post on X (formerly twitter) where a photo widely shared with misleading captions was used about a Hindu Priest performing a vedic prayer at Washington after recent elections. After investigating, we found that it shows a ritual performed by a Hindu priest at a private event in White House to bring an end to the Covid-19 Pandemic. Always verify claims before sharing.

Claim:

An image circulating after Donald Trump’s win in the US election shows Pujari Harish Brahmbhatt at the White House recently. 

‍

Fact Check: 

The analysis was carried out and found that the video is from an old post that was uploaded in May 2020. By doing a Reverse Image Search we were able to trace the sacred Vedic Shanti Path or peace prayer was recited by a Hindu priest in the Rose Garden of the White House on the occasion of National Day of Prayer Service with other religious leaders to pray for the health, safety and well-being of everyone affected by the coronavirus pandemic during those difficult days, and to bring an end to Covid-19 Pandemic. 

‍

Conclusion:

The viral claim mentioning that a Hindu priest performed a Vedic prayer at the White House during Donald Trump’s presidency isn’t true. The photo is actually from a private event in 2020 and provides misleading information.  

Before sharing viral posts, take a brief moment to verify the facts. Misinformation spreads quickly and it’s far better to rely on trusted fact-checking sources.

• Claim:  Hindu priest held a Vedic prayer at the White House under Trump
• Claimed On:Instagram and X (Formerly Known As Twitter)
• Fact Check: False and Misleading 

Introduction

Election misinformation poses a major threat to democratic processes all over the world. The rampant spread of misleading information intentionally (disinformation) and unintentionally (misinformation) during the election cycle can not only create grounds for voter confusion with ramifications on election results but also incite harassment, bullying, and even physical violence. The attack on the United States Capitol Building in Washington D.C., in 2021, is a classic example of this phenomenon, where the spread of dis/misinformation snowballed into riots. 

Election Dis/Misinformation 

Election dis/misinformation is false or misleading information that affects/influences public understanding of voting, candidates, and election integrity. The internet, particularly social media, is the foremost source of false information during elections. It hosts fabricated news articles, posts or messages containing incorrectly-captioned pictures and videos, fabricated websites, synthetic media and memes, and distorted truths or lies. In a recent example during the 2024 US elections, fake videos using the Federal Bureau of Investigation’s (FBI) insignia alleging voter fraud in collusion with a political party and claiming the threat of terrorist attacks were circulated. According to polling data collected by Brookings, false claims influenced how voters saw candidates and shaped opinions on major issues like the economy, immigration, and crime. It also impacted how they viewed the news media’s coverage of the candidates’ campaign. The shaping of public perceptions can thus, directly influence election outcomes. It can increase polarisation, affect the quality of democratic discourse, and cause disenfranchisement. From a broader perspective, pervasive and persistent misinformation during the electoral process also has the potential to erode public trust in democratic government institutions and destabilise social order in the long run.  

Challenges In Combating Dis/Misinformation

‍

1. Platform Limitations:  Current content moderation practices by social media companies struggle to identify and flag misinformation effectively. To address this, further adjustments are needed, including platform design improvements, algorithm changes, enhanced content moderation, and stronger regulations.

2. Speed and Spread: Due to increasingly powerful algorithms, the speed and scale at which misinformation can spread is unprecedented. In contrast, content moderation and fact-checking are reactive and are more time-consuming. Further, incendiary material, which is often the subject of fake news, tends to command higher emotional engagement and thus, spreads faster (virality). 

3. Geopolitical influences: Foreign actors seeking to benefit from the erosion of public trust in the USA present a challenge to the country's governance, administration and security machinery. In 2018, the federal jury indicted 11 Russian military officials for alleged computer hacking to gain access to files during the 2016 elections.  Similarly, Russian involvement in the 2024 federal elections has been alleged by high-ranking officials such as White House national security spokesman John Kirby, and Attorney General Merrick Garland. 

4. Lack of Targeted Plan to Combat Election Dis/Misinformation:  In the USA,  dis/misinformation is indirectly addressed through laws on commercial advertising, fraud, defamation, etc. At the state level, some laws such as Bills AB 730, AB 2655, AB 2839, and AB 2355 in California target election dis/misinformation. The federal and state governments criminalize false claims about election procedures, but the Constitution mandates  “breathing space” for protection from false statements within election speech. This makes it difficult for the government to regulate election-related falsities.

‍

‍

CyberPeace Recommendations

1. Strengthening Election Cybersecurity Infrastructure: To build public trust in the electoral process and its institutions, security measures such as updated data protection protocols, publicized audits of election results, encryption of voter data, etc. can be taken. In 2022, the federal legislative body of the USA  passed the Electoral Count Reform and Presidential Transition Improvement Act (ECRA), pushing reforms allowing only a state’s governor or designated executive official to submit official election results, preventing state legislatures from altering elector appointment rules after Election Day and making it more difficult for federal legislators to overturn election results. More investments can be made in training, scenario planning, and fact-checking for more robust mitigation of election-related malpractices online. 
2. Regulating Transparency on Social Media Platforms: Measures such as transparent labeling of election-related content and clear disclosure of political advertising to increase accountability can make it easier for voters to identify potential misinformation. This type of transparency is a necessary first step in the regulation of content on social media and is useful in providing disclosures, public reporting, and access to data for researchers. Regulatory support is also required in cases where popular platforms actively promote election misinformation.
3. Increasing focus on ‘Prebunking’ and Debunking Information: Rather than addressing misinformation after it spreads, ‘prebunking’ should serve as the primary defence to strengthen public resilience ahead of time. On the other hand, misinformation needs to be debunked repeatedly through trusted channels. Psychological inoculation techniques against dis/misinformation can be scaled to reach millions on social media through short videos or messages.
4. Focused Interventions On Contentious Themes By Social Media Platforms: As platforms prioritize user growth, the burden of verifying the accuracy of posts largely rests with users. To shoulder the responsibility of tackling false information, social media platforms can outline critical themes with large-scale impact such as anti-vax content, and either censor, ban, or tweak the recommendations algorithm to reduce exposure and weaken online echo chambers. 
5. Addressing Dis/Information through a Socio-Psychological Lens: Dis/misinformation and its impact on domains like health, education, economy, politics, etc. need to be understood through a psychological and sociological lens, apart from the technological one. A holistic understanding of the propagation of false information should inform digital literacy training in schools and public awareness campaigns to empower citizens to evaluate online information critically. 

‍

Conclusion

According to the World Economic Forum’s Global Risks Report 2024, the link between misleading or false information and societal unrest will be a focal point during elections in several major economies over the next two years. Democracies must employ a mixed approach of immediate tactical solutions, such as large-scale fact-checking and content labelling, and long-term evidence-backed countermeasures, such as digital literacy, to curb the spread and impact of dis/misinformation.

‍

Sources

•  https://www.cbsnews.com/news/2024-election-misinformation-fbi-fake-videos/ 
•  https://www.brookings.edu/articles/how-disinformation-defined-the-2024-election-narrative/ 
•  https://www.fbi.gov/wanted/cyber/russian-interference-in-2016-u-s-elections
• https://indianexpress.com/article/world/misinformation-spreads-fear-distrust-ahead-us-election-9652111/ 
•  https://academic.oup.com/ajcl/article/70/Supplement_1/i278/6597032#377629256 
• https://www.brennancenter.org/our-work/policy-solutions/how-states-can-prevent-election-subversion-2024-and-beyond 
•  https://www.bbc.com/news/articles/cx2dpj485nno 
• https://msutoday.msu.edu/news/2022/how-misinformation-and-disinformation-influence-elections 
• https://misinforeview.hks.harvard.edu/article/a-survey-of-expert-views-on-misinformation-definitions-determinants-solutions-and-future-of-the-field/ 
• https://reutersinstitute.politics.ox.ac.uk/sites/default/files/2023-06/Digital_News_Report_2023.pdf 
• https://www.weforum.org/stories/2024/03/disinformation-trust-ecosystem-experts-curb-it/ 
• https://www.apa.org/topics/journalism-facts/misinformation-recommendations 
• https://mythvsreality.eci.gov.in/ 
• https://www.brookings.edu/articles/transparency-is-essential-for-effective-social-media-regulation/ 
• https://www.brookings.edu/articles/how-should-social-media-platforms-combat-misinformation-and-hate-speech/ 

‍