From Hulku to Boltu: Inside India's AI slop economy

The Hulk is an unaccountably popular figure in the pantheon of characters that populate AI generated content in India. The Marvel superhero, who first appeared in 1962 in the comic The Incredible Hulk by Stan Lee and Jack Kirby, has been reborn in Indian AI content as “Hulku”, “Hulkeshwar”, and “Hulkanand”, and has featured in thousands of user-generated AI videos. He is usually a villager, often bullied by a frail Indian grandmother who orders him around and makes him do household chores. He is also teased by village bullies, scolded by elders, and asked why he’s not married.

So popular is the Hulk genre on platforms that many popular YouTube channels have viral video tutorials on making a Hulk video in three easy steps, including readymade lists of story-writing prompts that can be fed into ChatGPT to generate a script. Watching these videos, one guesses at the reasons behind their popularity—in one, the Hulk figure is thrown out of a mall for being poorly dressed—and we get glimpses of the fears and insecurities that challenge and motivate a large proportion of the Indian population. What looks like disposable AI “slop” is actually the most visible side of a fast-growing creator economy—one that platforms encourage, monetise, but, contradictorily, try to disown.

The most successful AI channel—not just in India but worldwide—is not a Hulk-based one (though Hulk does make a special appearance in its videos once in a while). It is Bandar Apna Dost, the YouTube channel that shot to fame in November 2025 when a widely shared global report stated that it was the “AI slop” channel with the most views in the world on YouTube—2.07 billion views at the time (2.65 billion views today). The anthropomorphised monkey Boltu is its main character and shares many traits with “Halku”—but maybe in a sea of Hulks, it stood out for being same-but-different.

AI Slop Report: The Global Rise of Low-Quality AI Videos by the San Francisco-based Kapwing, an online, AI-powered video creation and editing platform, also named Bandar Apna Dost as the channel with the highest estimated revenue—$4.25 million—a figure it arrived at by estimating yearly revenues of the top trending YouTube channels in every country using social media analytics tool Social Blade. Surajit Karmakar, the 27-year-old content creator behind Bandar Apna Dost, does not either confirm or deny the number. “I’m not able to disclose my annual income figures, but yes, I do earn very well,” he told Lounge.

The creator's journey

From Bongaigaon, Assam, Karmakar’s journey as a content creator started as a maker of technical videos back in 2014. His father had left the family when he was in high school, and he wanted to help his mother run the household, made up of him and his sister. The tech videos channel didn’t quite take off, and Karmakar switched to making comedy and “love story” videos on his channel Besharam Boyz with a bunch of friends. “We would write a script, hire actors, do 2-3 days of shooting—it was a lot of work,” says Karmakar. The channel was fairly successful—it had 1.5 million subscribers on YouTube. Then the pandemic struck, Karmakar had to stop shooting and his uploads dwindled. Then, his channel got multiple copyright strikes for music and was shut down by YouTube. “This was in July 2024, and we had been creating content since 2017… it felt like my YouTube journey had come to an end,” says Karmakar.

He discovered AI video content made using Google’s Veo 3 in 2024—and within that, animal videos doing well. “I saw that most creators were not consistent—they would upload a few videos, one or two would go viral, and then they would stop. I wanted to see how far I could take it—initially, I uploaded 20-25 bandar videos a day,” says Karmakar. “I thought it was great that I could work totally on my own. I didn’t need a team, a cameraman, actors—this was just me.” Within a month, he had hit 100,000 subscribers and had started getting monetised. Today, Karmakar takes more time over his videos—it takes him around 6-7 hours to make one—and uploads one video a day on his YouTube and Instagram pages (he has 121,000 followers on Instagram). “People call these low-effort videos, junk food, slop, etc., but I want them to know that I put in a lot of work into this. Please show me how to make these videos in 5-10 minutes.”

In late 2024, tech journalist Casey Newton coined the term “AI slop” to describe the flood of quickly produced, AI-generated content quietly overwhelming the internet. In June 2025, John Oliver did a segment on the phenomenon—naming India as one of the top producers of this kind of content. “AI slop” was named word of the year by Merriam-Webster dictionary in 2025. It is associated with bizarre, fantastical content, often featuring anthropomorphised animals and hyper-emotional narratives.

There has been a sort of moral panic over AI slop and what it is doing to the internet—even Kapwing’s study calls it “slop” and “low-quality AI videos”. But for creators, it is a content-generation engine, and following Karmakar’s success, perhaps even a wealth-generation one. Many creators who worked with live action, short-form vertical videos, which were the dominant form of mainstream entertainment on social media for years, and which took days and weeks to produce, have pivoted to using AI to generate videos in a much shorter time and increase their earnings.

Bengaluru-based Prajwal Rai is one of them. Rai, 32, had been a video editor and ad filmmaker, running a small studio where he would make ads for local companies and brands, for several years when he discovered AI video generation tools. He had tried his hand at being a social media content creator making short scripted videos with actors, but without much success. Today, after a decade of studio work, Rai has pivoted almost entirely to being an AI content creator.

“AI has totally changed the game. My page is not the main revenue earner for me—that’s mainly to showcase my capabilities. The main revenue comes from making AI-generated content for clients, for ads, personal events, etc.,” says Rai. AI-generated films for landmark events like weddings, 50th birthdays and anniversaries are a huge trend now, says Rai. “The family shares a few photos and sound bytes, and then we create the rest of the narrative using AI. They are shown at the event, and people love them.” Rai charges around ₹1 lakh for each such video, and his earnings per month have exceeded ₹3 lakh in good months. He has also written and self-published a book called AI Me: A Simple Guide to AI and Saving Your Job.

Over the past year, the AI content ecosystem in India has grown to encompass creators, platforms, and content coaches who teach people how to use AI tools to generate videos. Some run courses on Udemy and Coursera, while many have their own YouTube channels and Instagram pages where they regularly publish videos on prompt engineering and comparisons between different AI content creation platforms like Kling AI, Sora, Veo 3, Luma Dream Machine, and Higgsfield AI. Meanwhile, Indian AI-powered video-generation platforms like Invideo and Phenomenal AI are also snapping at the heels of the established global players.

One of the latest videos on Instagram and AI content coach Tanishaa Bhansali’s Instagram page (which has 318,000 followers) is about creating a trending AI video that shows the user riding in a car with Hindu gods. In a 30-second video, Chennai-based Bhansali, 25, teaches her viewers how to create a similar video featuring them in the driver’s seat. Bhansali has coached over 40,000 people—and in the past year, she has seen a huge rise in interest in making AI-generated videos. Over the past couple of months, AI cloning has emerged as one of the big ways of using AI tools like HeyGen and Eleven Labs, she says—from influencers to famous podcasters, everyone is creating AI clones of themselves, which replicate their physical form, voice and mannerisms that can then churn out even more videos without the creator having to actually shoot them in front of a camera.

The quality of AI-generated videos is always improving, says Bhansali. “Every month you will see new changes and advances and more realistic content. Of course, you need to know how to prompt the right way, and one of the most important things is negative prompting. People usually say ‘I want you to create this, this and this’ but what they forget is to tell AI what not to do as well—saying things like ‘make sure you do not change my face’, ‘do not change my hair colour, my skin texture or make it look plastic’,” explains Bhansali.

The future of AI slop

Like most forms of entertainment, there are trends and variations in “AI slop”, and as with most things AI, the evolution is very fast. Formats and templates rise to the top of the charts almost every day, and are endlessly replicated in different contexts. Currently, there seems to be a rise of videos that combine live-action with AI-generated animation, such as a human interacting with an AI character — a popular video shows a house cat playing music in the middle of the night and being scolded by its owner. “This is a fast evolving space, still in a very initial native space. Ultimately, it is storytelling, and we have been using various tools for storytelling forever. It is a bit elitist to call these videos ‘slop’, because the emotions they depict don’t appeal to you. I see a massive opportunity, where people who have a storytelling skill set, but didn’t have the tools to do it till now, are going to be able to do that,” says Himanshu Arora, co-founder, Social Panga, a digital marketing agency based in Bengaluru.

Dipankar Mukherjee, a former ad executive who runs Mumbai-based StudioBlo, an AI-native filmmaking company, thinks of AI as a creative and cultural tool, not a gimmick. He insists that AI-led storytelling must also wrestle with deeper questions around authorship, consent, ethics, and ownership. “I think AI slop can be broadly broken down into two categories—one is senseless, mindless automation at scale, which will die its natural death and also be killed by Meta and Google itself,” says Mukherjee, referring to the platforms’ ever-evolving rules and regulations around the monetization of AI content. “The other is content made with humour and intent, which eventually will get more sophisticated. The wild, wild west of it will die down and people will figure out a way to monetise and legalise it. Think of memes or GIFs as cultural artefacts—they are also built on top of existing IP, but recontextualised, right? AI slop is similar—it can be produced more quickly, but parts of it will remain in the public consciousness or become a part of pop culture only if they are good enough,” says Mukherjee.

There’s also a growing—and slightly hypocritical—tension at the heart of creator platforms like Instagram and YouTube: they’re aggressively pushing AI tools, while simultaneously tightening the screws on monetisation when creators actually use them at scale. Instagram encourages creators to use generative AI stickers, AI backgrounds, auto-captioning and remix tools to “create faster and post more,” while YouTube promotes Dream Screen, AI-generated backgrounds, auto-dubbing and experimental generative video features inside Shorts. Yet, both platforms have become far stricter about what qualifies for monetisation. In July 2025, YouTube updated its monetisation policy, renaming the “repetitious content” policy to “inauthentic content”. While AI tools are permitted for creation, videos must be original, high-value, and not easily replicated at scale to qualify for the YouTube Partner Program (YPP), the company said.

There is a lot of opacity around what qualifies as “high value” content and what is rubbished as “low quality”. Currently, it seems to be “you know it when you see it”, which is not quite a fair metric. It is clear that AI slop is going to be a big concern for platforms in 2026 as they deal with vast amounts of content that cannot be clearly and succinctly categorised. There is suddenly an emphasis on “rawness” and “authenticity”, which have acquired a tinge of being morally better than “inauthentic” AI content. Three weeks ago, Adam Mosseri, head of Instagram, shared a post: “A few thoughts on why creators will continue to matter, how aesthetics are changing, and no longer believing what we see in a world of infinite synthetic content as we head into 2026. People want content that feels real. We are going to see a significant acceleration of a more raw aesthetic over the next few years. Savvy creators are going to lean into explicitly unproduced and unflattering images of themselves. In a world where everything can be perfected, imperfection becomes a signal,” he wrote.

Will India’s content creators—a tough, adaptable breed, as Karmakar and Rai show us—be able to compete in this scenario? Not if the ruling aesthetic is defined by a privileged few whose language is global, not local.