Political misinformation keeps India’s fact-checkers in overdrive

NEW DELHI – When Indian opposition leader Rahul Gandhi was seen making a stunning claim in a speech being circulated on social media, Ms Pallavi Mishra knew it was time for her to get to work.

Seated in her office on the 20th floor of a tower that overlooks the vast urban sprawl of Noida, a city adjacent to the Indian capital Delhi, she pulled up an Instagram video of the speech on her laptop.

In the video, Mr Gandhi says: “Let me tell you the truth right at the start: 2024, June 4, Narendra Modi will remain the prime minister of India.”

June 4 is the day when the results of the ongoing Indian general election will be declared.

A seasoned fact-checker, the 35-year-old Ms Mishra knew the video – which had been sent to the WhatsApp account of Vishvas News, the fact-checking organisation she works for – was fake the moment she saw it. But she had to find proof to label it false.

A reverse image search of a screenshot of Mr Gandhi in the fake video – where he is seen in a white T-shirt and foregrounded by a unique flower arrangement on a lectern – quickly led her to the original video of his speech online.

In that, Mr Gandhi is saying the exact opposite, that come “2024, June 4, Narendra Modi will not remain the prime minister of India”.

By editing out a single word, the fake clip attempted to give the impression that Mr Gandhi and his Congress party had conceded defeat, adding to the narrative of a weak Indian opposition.

Since India’s mammoth seven-phase election began on April 19, fact-checkers such as Ms Mishra have been busy trying to stem an online surge of misinformation aimed at influencing

the country’s nearly 969 million registered voters.

Claims circulating online

– backed by fake graphics – have predicted exit poll wins for specific parties, while others have stated without evidence that the election will be rigged.

Top Bollywood actors were not spared, appearing in deepfake videos that show them criticising Mr Modi and seeking votes for the Congress party. Other posts seek to incite hatred against India’s Muslims in the polarised election environment.

But, contrary to expectations that more advanced deepfakes would wreak havoc on the electoral process, it is “cheapfakes” – content developed with little effort by using simple video editing or recontextualising authentic images or videos to deliberately alter their meaning – that have kept Indian fact-checkers busy.

Since the last general election in 2019, the number of Indian internet users has surged to more than 850 million. However, digital literacy has lagged behind, aggravating risks from such political disinformation.

According to a World Economic Forum survey released in January, India faces a higher risk from misinformation than infectious diseases or illicit economic activity in the next two years.

“Many (in India) now have access to smartphones and the internet without the awareness to use them judiciously,” said Ms Ruby Dhingra, managing editor of Newschecker.in, a fact-checking portal. “There are a lot of people who have grown up thinking that seeing is believing, so they are not even suspicious of what they see online.”

Tech companies such as Meta and Google have also been accused of failing to implement adequate measures to limit the spread of harmful content that risks distorting public opinion or provoking violence, including disinformation and hate speech, on their platforms.

Prior to the election, Meta in April expanded its third-party fact-checking network in India. This now includes 12 partners covering 17 languages including English, making it the tech giant’s largest such network for a country.

And together with the Misinformation Combat Alliance, a cross-industry alliance to combat fake news, Meta also launched a dedicated fact-checking helpline on WhatsApp to combat artificial intelligence-generated misinformation in India.

Before the election, Google too announced its support for Shakti: India Election Fact-Checking Collective, a consortium of news publishers and fact-checkers who are collaborating to detect online misinformation early and amplify the reach of fact-checks in Indian languages.

The problem of misinformation in India is compounded by the rise of influencers, many of whom hold sway and have credibility with their hundreds of thousands of followers. And when fake news – even if old – is regurgitated by such actors, it poses renewed risk.

“The claims could easily be the same, but then when the reach increases, you know the impact is bigger,” said Ms Mishra. “Our fact-checks have therefore become more relevant and more important. Even if we are reproducing an old fact-check, it has value.”

All the Indian fact-checking organisations mentioned in this story are part of Meta’s global third-party fact-checking network. They work with Meta to identify misinformation on its platforms so that the company can restrict the reach of such harmful content.

They are also signatories to the Code of Principles of the International Fact-Checking Network, which was launched in 2015 by Poynter, a non-profit media institute in the US, and have their independent websites and social media accounts where they feature their work. But the reach of their fact-checks remains limited compared with the viral extent of misinformation.

Less than 20km from Ms Mishra’s office, in an upscale leafy neighbourhood in Delhi, where a residential bungalow has been converted into multiple offices, Mr Archis Chowdhury, 32, a senior correspondent with Boom, another fact-checking organisation, pulls up some of the false claims he has debunked this election season.

One of them pertained to a fake image of the front page of a leading Hindi daily, Dainik Bhaskar, in which the lead story claimed a survey had shown that the Congress-led opposition alliance was leading in 10 states.

Other posts alleged the Congress party had not fielded any candidate in the ongoing election in the north-eastern state of Arunachal Pradesh, which is claimed by China as its territory, because Beijing “might get angry”. They also claimed it would breach a “2009 memorandum of understanding signed between the Congress and the Chinese Communist Party”.

Mr Chowdhury had seen such posts even prior to the election. But this time, he has observed a new trend where even official state-level social media accounts of political parties peddle fake news.

An example was the Congress’ Jharkhand state unit posting a fake video on X showing India’s Home Minister Amit Shah claiming that the Bharatiya Janata Party would abolish affirmative action measures for marginalised caste groups. The @INCJharkhand account on X was suspended in response to a legal demand from the authorities.

The fake video featuring Mr Shah was doctored with basic editing tools, one of the many so-called “cheapfakes” that have dominated misinformation in India.

Another such cheapfake was put out by official X accounts of the Bharatiya Janata Party in West Bengal and Tripura to showcase the development of metro railway services under its rule, but they did so using an image of Singapore’s Jurong East MRT station.

Since April, India has witnessed an intense political slugfest, with false claims delivered by even top political leaders, unleashing a surge in misinformation backing their statements.

For instance, when Mr Modi falsely claimed in an election rally in April that the Congress party would confiscate the assets of Indians and distribute them to “those with more kids” – a veiled and stereotypical reference to Muslims in India – there was “a whole surge in Islamophobic posts and false claims”, said Mr Chowdhury.

Such claims go viral as they “trigger an emotional response in people that makes them share it”, he said.

In contrast, “a fact-check relies on logic, critical thinking, which makes it less shareable just by its nature”, he added.

Despite safeguards put in place by tech companies before the election, shortcomings were exposed. Meta, which owns Facebook and Instagram, was found to have approved AI-manipulated political adverts that sought to spread disinformation and incite religious violence.

The Guardian newspaper reported on May 20 that these adverts contained slurs towards Indian Muslims, such as “let’s burn this vermin”, as well as disinformation about political leaders.

The adverts were created and submitted to Meta’s ad library by India Civil Watch International and Eko, a corporate accountability organisation, to test the company’s mechanisms for detecting and blocking harmful political content during the ongoing election.

Fact-checkers ST spoke with also said that tech companies could further expand their fact-checking networks, as well as invest in equipping fact-checkers with better resources to match the rapid strides in AI and other technologies.

And, given that fact-checks are “not as catchy or engaging” as the original misinformation, Ms Dhingra suggested that tech companies’ algorithms could promote fact-checking content put out by organisations like hers to amplify the reach of the work that fact-checkers do. “If they could do that and ensure more people see it, then that could maybe help.”

A Meta spokesperson told ST that the company offers “regular training to our fact-checking partners and also provides them access to critical tools that help with detection and verification to better address misinformation online”.

Meta in November 2023 had provided its fact-checkers with access to the Meta Content Library and tools that give them “the most comprehensive access to publicly available content across Facebook and Instagram” so that they can investigate and debunk misinformation more effectively.

While the exact impact of misinformation on Indians’ voting preferences is unclear, Dr Joyojeet Pal, an associate professor of information at the University of Michigan, said that such content sharpens existing political biases.

When there is misinformation around Mr Modi or Mr Gandhi, it acts at a national level, he said, affecting not only the public persona of either leader, but also impacting “the way people perceive voting for that party”.

“When the misinformation is of a micro level – say a local influencer making claims about a politician – that is much more impactful at the level of the specific candidate and their prospects,” Dr Pal added.

At the heart of the misinformation problem is an individual’s confirmation bias – the tendency to search for and favour information that affirms one’s prior beliefs or values – which is difficult to unseat.

It is a challenge Ms Urvashi Kapoor, a senior editor with Vishvas News, often faces in the fact-checking organisation’s public digital literacy workshops.

“When we conduct training, we often find that some people are resistant to reasoning,” she said. “Those who blindly follow a particular party tend to follow its leaders unquestioningly, even when misinformation is involved.”