How AI Is Helping Combat Misinformation in India
The internet and unregulated social media platforms have heightened the threat that misinformation and disinformation pose to us individually, nationally, and globally. Mis/disinformation is a borderless problem, but it’s especially acute in India. With over 650 Million internet users in the largest democracy in the world, domestic and foreign actors utilize disinformation as a weapon - creating fresh challenges for Indian legislators, law enforcement officials, and citizens.
Cybercrime continues to cause real-world harm as the relationship between digital and physical spaces becomes increasingly interlinked. Malign actors use innovative technology and data gathered from open sources to spread misleading information to further their agendas. They are able to manipulate the algorithms that power social media and search engine platforms, allowing fake news and harmful disinformation to gain considerable visibility. The anonymous nature of the internet, the virality of content, and the spectrum of techniques being employed by bad actors mean that identifying and combating misinformation and disinformation is very challenging.
Combating misinformation and cybercrime with technology
The spread of misleading information is particularly prevalent in India. It can damage company and individual reputations, disturb the peace, and have fatal consequences. Some of the most harmful rumors are based on sensationalist content that evokes strong emotions. Technology gives us the means to share and edit content within a large group leading to false rumours reaching people before the truth. This has resulted in community conflicts, misunderstandings and tension between different religious groups. Mob lynching is one of the drastic outcomes of rumors being spread on social media platforms and closed messaging groups.
Rigorous, evidence-based fact checks are a powerful tool in stopping the spread of harmful rumors. However, searching for verified information, evaluating its credibility, and writing fact-checking reports is a time-consuming process. There’s also the added complexity of multimedia misinformation such as memes, deep fakes and imagery being taken out of context.
Fortunately, AI can expedite fact-checking processes. Logically has developed a fact matching model built on thousands of fact checks to reference previously verified information. We have also developed an evidence retrieval system that is able to find pertinent information relating to fact-check claims from credible sources as well as a host of other systems, including vision AI for dealing with manipulated/fabricated multimedia content.
Anil Bandakhavi, Head of Data Science at Logically, said:
“Building automated tools for fact-checkers hasn't been an easy feat. We’ve trained and deployed advanced NLP models to understand and contextualize segments of text as well as the text as a whole. This is then cross-referenced with our vast knowledge base of credible sources and fact checks to power our evidence retrieval and fact matching tools, allowing us to speed up the fact-checking process and ensure we’re not doubling up on work.”
These technologies enable us to debunk rumors in a very short time frame. Some examples are:
- A video of Muslims protesting against Haridwar Dharam Sansad's hate speech in Bareilly on January 2022 has been falsely linked to the Prophet Muhammad row.
- A video of clashes between police and BJP workers outside Bhadrak Collector's Office, January 2022, has been misattributed.
- False claim that the temples in Tamil Nadu have to pay more for electricity than churches or mosques.
- False claim that Congress MP Rahul Gandhi called Udaipur killers "children”.
- False claim that Sandeep Chaudhary was injured during a protest against Kanhaiya Lal's murder in Rajsamand. The accused does not belong to the Muslim community.
Unfortunately, fact-checking is often not enough to dissuade those who have already subscribed to a belief based on misinformation and it will never be a one-size-fits-all solution. We cannot guarantee that fact-checks debunking misinformation will be engaged with. In some cases, deeper investigation into those pushing false narratives need to be conducted to establish motives of bad actors and understand how they operate in online spaces.
Logically’s report into conspiracy blog-style news website Great Game India (GGI) shows how rumours can spread on a global scale. GGI originally published an article detailing a popular COVID-19 conspiracy theory which has been re-published by a number of high-readership websites globally. Learn more by viewing the report here.
Responsibility, Explainability and Ethics in AI
Due to the power of AI and its potential to have a profound impact on people’s lives, this technology needs to be developed with care. Learn about using AI responsibly and ethically.
How AI Helps Us Fact-Check in a Crisis
How can AI help during a crisis? Learn how we built an AI pipeline to help us source fact-checkable claims during the Russia-Ukraine war.
3 Examples of OSINT Techniques Used To Expose Chinese Operations
Find out how the relatively new phenomenon of deepfakes work.