West Bengal elections: How parties are using AI to shape narratives

New Delhi,UPDATED: Apr 23, 2026 16:56 IST

West Bengal has long been known for fiercely contested elections. That intensity has only grown; only the rivals have changed. Today, the battle is between the All India Trinamool Congress and the Bharatiya Janata Party.

And as digital media increasingly shapes political narratives, the hostility between the ruling party and the challenger seeking to break into one of its last major strongholds has now spilled into the world of AI.

The Trinamool Congress have taken the lead in AI-generated content on its Instagram pages. The BJP, however, has posted some of the most overt AI-driven content, closely aligned with its broader political messaging.

Picture this: a photorealistic social media reel shows the Trinamool Congress’s top leaders huddled inside a jail cell.

Another sequence shows a group of strongmen harassing businesses and looting rations from public distribution shops. Their religious identity is never explicitly identified in the voice-over. Yet the visual coding: all of them wearing skullcaps, may be read as signalling a religious identity.

Many such scenes may have never happened. Yet, through AI, they appear believable enough to shape political narrative.

As West Bengal heads to the polls, political parties are using generative AI on a never-before-seen scale and scope, an analysis by India Today’s Open-Source Intelligence (OSINT) team suggests.

We reviewed roughly 1,500 Instagram reels from the official accounts of BJP, Trinamool, and CPI(M). The last 500 posts from each party, published up to April 18, were examined.

The All India Trinamool Congress (AITC) posted at least 56 short videos that were fully or partly created using AI, followed closely by the BJP with 43. CPI(M) published 25 such reels, according to data collected through Apify’s Instagram Scraper.

India Today found that the ruling party is not only leading in the use of AI-generated content, but also posting more aggressively on Instagram. AITC uploaded the roughly 500 reels analysed here between March 21 and April 18 in the shortest span of just 29 days. The BJP posted a similar volume between March 4 and April 18, a span of 46 days, while CPI(M)’s uploaded dataset stretches from December 9 to April 18, covering 131 days.

Why AI?

The analysis of the 1500 posts by India Today shows that one in every nine reels posted by AITC was made using AI. For the BJP, roughly every 12th post uses AI; for the CPI(M), it is about every 20th post.

But why are parties leaning more towards AI content? The answer is simple. Because the audience seems to like them more.

Across the 124 AI reels posted by these three parties, one pattern is hard to miss: AI content drew far more attention. India Today’s analysis of scraped Instagram data found that the average AI reel received about 6,600 likes, against roughly 3,500 for a non-AI reel, indicating that AI videos were drawing nearly double the engagement.

Forgotten cartoons, revived by AI

For many Bengalis, Thakurmaa, the grandmotherly storyteller from Thakurmar Jhuli, is more than just a cartoon character; she is a cherished piece of childhood nostalgia. Now, that familiar face has returned, not to television, but to the Instagram pages of political parties during election season, telling viewers whom to vote for.

On the BJP’s official Instagram handle, the much-loved Thakurmaa returns not as a keeper of folklore, but as a political narrator. In these AI-generated retellings, Bengal’s past is recast into campaign messaging, with Chief Minister Mamata Banerjee portrayed as a menacing queen blamed for stealing people’s happiness and dragging the state into darkness.

The Trinamool, too, has turned to AI caricature. Its reels recast BJP leaders in mocking avatars: Prime Minister Narendra Modi appears in an AI-cartoon avatar, while Home Minister Amit Shah is depicted in a demeaning mouse costume.

While the BJP creatively drew on familiar cultural figures such as Goopy Gyne Bagha Byne, Mugambo, Byomkesh Bakshi and Thakurmar Jhuli to sharpen its political storytelling, CPI(M) turned to characters like Chhota Bheem and Shaktimaan. The Trinamool, by contrast, relied less on nostalgic icons and more on fictional AI-generated avatars of top BJP leaders.

Digital Reconstruction of Tragedy

Another recurring theme across many AI-generated reels is the digital reconstruction of deeply sensitive tragedies - from the factory fire at a Wow! Momo outlet that killed workers to the R.G. Kar rape case. Yet not all of these videos are rooted in fact.

In one instance, an AI-generated video appeared to recreate the reported killing of two Dalit students during protests, allegedly at the hands of the West Bengal police. The same reel depicts people presented as Urdu teachers spitting a red substance on school walls and floors. By blending real events with provocative visual suggestion, such content deepens existing social fault lines.

At the same time, the ruling party is using AI to stage speculative and entirely fabricated visions of a future under BJP rule in their AI-generated reel series “What if they come?”. In some of these purely imagined sequences, even symbols of Bengali cultural identity, such as a statue of Rabindranath Tagore, are shown being vandalised or destroyed.

Mudslinging and personal attacks are not new to Indian elections. But until now, there were limits to what parties could convincingly produce and distribute.

What’s the damage?

During the Bihar elections, India Today’s analysis of AI-generated political content showed the BJP as the clear frontrunner. This time, it is running almost neck and neck with Bengal’s ruling Trinamool Congress.

What makes many of these reels especially concerning is that they often carry no clear AI label, even when they present hyper-realistic versions of events. For less-literate audiences, such videos can still appear trustworthy. In many remote parts of Bengal, the idea of “synthetic” or AI-generated content is unfamiliar.

For ordinary posts and reels on Instagram, Meta says people must use its AI-disclosure tool when they post photorealistic video or realistic-sounding audio that was digitally created or altered and could materially mislead people on a matter of public importance. Meta also says it may add AI info labels to a wider range of videos, audio and images when it detects industry-standard AI markers or when users disclose the content themselves.

- Ends

Tune In