How AI has invaded Indian cinema

On 3 January, the trailer for Vijay’s Jana Nayagan dropped. Amidst the fan adulation, especially feverish since this is Vijay’s last film before he turns to politics, a number of viewers spotted something amiss at the 23-second mark, a shot of Vijay with a rifle. In the lower right corner of the screen was a watermark of Google Gemini, an AI tool. The watermark has since been removed and it remains unclear how AI was used to enhance the trailer, or the final film, which is stuck in an ongoing battle with the censor board. Yet, its very presence speaks to the rapid rise of AI in Indian filmmaking over the last couple of years and the uncertainty around how it’s being used.

There are wildly varying opinions on AI in filmmaking. It’s a shiny new toy. It’s a useful tool that can save time and money. It limits art. It expands cinema. It’s a slippery slope that collapses creativity and limits the need for imagination. It can help create what we’ve never seen on screen before. It’s not art if it emerges from a prompt. It is art if it moves you, regardless of how it’s created. It’s not a film if 300 people on a film set are replaced by 30 typing furiously into a computer. It’s actually not that simple, and those 30 people are themselves artists guiding stories into being. It’s the end of storytelling as we know it. It’s the future.

Wherever you fall on the debate, what’s clear is this is only the beginning. Conversations with several Indian filmmakers, writers and producers confirm that AI has already begun retooling workflows and reshaping the filmmaking process, from screenwriting to casting, shooting and post-production.

Things are moving so swiftly now that there’s fresh AI-related film news every week. In the last few days alone, Ajay Devgn announced Bal Tanhaji, an AI-powered prequel to his 2020 film Tanhaji: The Unsung Hero. Multiple projects are vying for the apparently coveted title of “India’s first AI film”. Last March, the trailer for Vivek Anchalia’s Naisha dropped, calling itself “India’s first AI-powered Bollywood film”. The romantic drama was supposed to release in May, but Anchalia says technology has advanced too swiftly and the film needs reworking. May saw the release of Love You, a Kannada feature film directed by S. Narasimhamurthy, marketed as the “world’s first AI Kannada movie”.

In August came the announcement of Chiranjeevi Hanuman—The Eternal, described as “India’s first AI-generated Hindi feature film”, due for release this year. The soundtrack will feature music by Trilok, the “world’s first AI-powered band”. The film is co-produced by Collective Media Networks as part of their “Historyverse”, which includes a massive slate (four films, three shows) of AI projects releasing over the next two years. The first Historyverse title came out in October with JioHotstar’s Mahabharat: Ek Dharmayudh, a gaudy 10-episode “animated” series with that uniquely off-putting AI aesthetic which has invaded our collective Instagram algorithms. The dialogue doesn’t always sync with the awkward mouth movements, faces don’t emote, frames don’t flow into scenes.

According to media analytics firm Ormax Media, which estimates streaming title viewership across platforms, the series was the most watched fiction series in the week it released, clocking 1.6 million views. Also reportedly aiming for a 2026 release is Vayuputra, a “devotional” animated Telugu feature centered on Hanuman (AI-enabled mythological projects appear to be one of the subgenres that will flood our screens in the next few years).

Bigger names also seem willing to bet on AI. Last year, Ajay Devgn announced Prismix, a new media company specialising in generative AI storytelling. Bal Tanhaji is the company’s first announced title. Director Shakun Batra has also given interviews about how he’s starting an AI division of his production house, Jouska Films.

In October, the first-ever Mumbai AI Film Festival was held. When the festival was announced, a Twitter/X spat ensued between the defensive young organisers (all from startup, not filmmaking backgrounds) and those who are more apprehensive about “AI filmmaking”. The jury and guests at the festival included Batra, comedian Tanmay Bhat and filmmaker Karan Anshuman. The winner for “Best Storytelling” was The Age Of Rage, a 4-minute short about an unlikely global pandemic that starts eliminating the world’s population. Guided by a voiceover, the film is packaged like a short-documentary-meets-well-packaged-YouTube-conspiracy-video, with flashy, frenetic visual aids to add impact to the voiceover, rather than a visually cohesive narrative with scenes and characters.

AI enthusiasts aren’t just looking ahead—they’ve also set their sights on films from the past. In August, Eros International re-released the Tamil version of Raanjhanaa with an AI-generated alternate ending in which Dhanush’s character survives, thereby rewriting the film’s original tragic climax. Director Aanand L. Rai wasn’t consulted nor informed, and called it “a betrayal” and a “reckless and dystopian experiment”. Eros Media Group chief executive Pradeep Dwivedi defended the decision, telling The Guardian that the altered ending was an “exploratory baby step” and that the company was “significantly evaluating” its library of close to 3,000 titles for similar AI treatments.

One of the issues behind the actor’s and writer’s strikes in the US in 2023 was a demand for protection from AI. There continues to be heated debate over AI in Hollywood, but little, if anything, here. Indian cinema, it seems, is content to wait and see how AI pans out.

SCREENWRITING IN THE AGE OF CHATGPT

“Written by ChatGPT” is a common slur these days, hurled at films and shows with generic, stale writing and dialogue. But how is the service actually being used within the writing process? Multiple screenwriters I spoke to insist that ChatGPT is great for research and generating visual aids to help bring pitches to life through visuals of the world and character mock-ups.

Writers also say it serves as a great assistant—something to bounce ideas off, keep track of different story elements, help identify plot holes and refine scripts and pitches. “If I write a synopsis or if I’m building a character, I can actually use ChatGPT to ask me questions so that I can delve deeper into the character, and identify gaps,” says screenwriter Sulagna Chatterjee (Feels Like Ishq, Adulting). “Half the dialogue feedback it gives is often bizarre and just dry and not how human beings speak at all. But the other structural level feedback that it gives is often pretty interesting.”

The reliance on sending material to fellow writers for feedback also appears to be dwindling in a ChatGPT world. “Peers sometimes can take four-five days to get back to you with feedback—and it’s mostly broad feedback. ChatGPT can get molecular in no time about things like scene transitions and the rhythm of the scene,” says one screenwriter, who preferred to remain anonymous.

Chatterjee says ChatGPT is even useful to assess pitches from the viewpoint of specific production houses. “I remember when I had written a first draft, I was like (to ChatGPT): ‘Look at this entire document, but from the viewpoint of the commissioning team at a certain production house or platform, based on the kind of stuff that they have commissioned in the past six months or one year’. Or ‘these are my four concepts that I have developed. Tell me in terms of priority which one I should pitch first.’ And it gave me a very comprehensive answer.”

But isn’t the chatbot’s relentlessly validating vibe antithetical to the constantly iterative creative process? Chatterjee says she found a specific prompt that helps it be more “balanced” in its feedback. “The reason why AI is scary and absolutely not something that should be completely depended on is because it’s trained to create that sense of validation. You have to train your AI to be more like a critic rather than just a ‘yas queen’ person.”

Several writers also say that ChatGPT fits the current environment of uncertainty well, where much of what producers are looking for is formulaic and “algorithmic” and expected timelines are tighter than ever. “Platforms today expect you to come with everything fleshed out, and the expected turnaround time is insane. One platform asked me to turn around a pilot episode for a show in two-three days. Then I don’t have a choice but to work with ChatGPT…I don’t love it. It’s just where we are,” says one writer. “There’s me in there somewhere. The voice is still mine. I’m just taking a little extra help because otherwise I have no other option. It’s either finish it off in whatever bizarre timelines you get or realise you are so replaceable,” offers another, adding, “But I think there’s still a certain level of internalised shame associated with using AI for screenwriting. There are times I’m sitting at a cafe and I will see a lot of writers there with multiple ChatGPT windows open.”

Writers also say certain producers have started sending out AI-generated plot outlines and telling screenwriters to build on those and essentially fill in the gaps. Not to mention the common belief that many producers are using ChatGPT to read, assess and summarise pitches rather than reading them themselves. As to whether AI might further diminish the value attached to screenwriters and screenwriting, Colour Yellow Movies COO Harini Lakshminarayan, who has been very vocal about the dangers of AI following the Raanjhana re-release, says it’s already happening. “I know more writers who have quit the industry in the last six months than I have ever seen and I’ve been around for 23 plus years. I have never known so many people to quit filmmaking. And it’s not just AI, it is the challenges of the economy, the industry, the mergers and acquisitions that are happening. But all of that is happening because there is this technology that people assume helps you fill the gap.”

“I’m a little scared and sad about 20-somethings who are just joining the industry for whom this is the new normal. And because the work is so little, the focus can become just churning things out like a factory with the help of AI,” says Chatterjee. One writer jokes that in the near future, “successful writers” will be those who specialise in taking ChatGPT-assisted writing and “unpolishing” it to make it sound more human.

But how has the ground shifted for those on the other side, on the receiving end of pitches and scripts? “You need to be well-versed with AI because you need to be able to recognise it when you see it,” says a development executive at a top Bollywood production house. She says “about 70%” of the pitches that come across her desk have some component of AI and it’s an immediate turn-off. “When I see ChatGPT pitches, I get damn ticked off… I miss bad pitches written by actual people.” But one benefit, she says, is that it’s an equaliser for writers for whom English isn’t the first language, given that a great deal of the pitch submission ecosystem is in English. So now, a writer with a great idea isn’t held back by language. She adds, however, that these are rare instances.

AI WILL FIX IT IN POST

The same development executive says AI is also “massively helpful” in creating concept art, storyboarding, pre-visualisation and lookbooks to help aesthetically align with and brief the various department heads. Production designers, cinematographers and costume designers increasingly use it to help bring characters and worlds to life. AI is even used in casting. When projects are pitched to top actors, they now routinely receive mock-ups of themselves as the character to help excite them about the role. In August last year, ahead of the release of the Ayushmann Khurana vampire comedy Thamma—part of the massively successful Maddock Horrorverse—director Aditya Sarpotdar told The Hollywood Reporter India that he used AI visuals as a reference to brief his stunt and VFX team on how he wanted the vampire characters to look when running and gliding across rooftops in that distinct “vampire way”.

Speaking to British Cinematographer magazine, cinematographer Ravi Varman (Ponniyan Selvan, Barfi) was remarkably frank about their use of AI in Shankar’s Indian 2 (2025). “This movie was shot over six years, with production halts due to covid. AI technology also allowed us to complete unavailable actors’ on-screen presence by using their body doubles and replacing their faces with realistic AI-generated faces… We also had to use digital de-ageing technology to reduce the age of the lead actor (Kamal Haasan) from 69 to around 30”. Varman, one of the best-regarded cinematographers working today, also zeroed in on why AI would only become more common in a country obsessed with making bigger, showier action films. “Commercial cinema often demands that actors perform superhuman feats, regardless of their physical limitations. Technologies such as virtual production, motion-controlled cameras, and AI have become essential tools in achieving these expectations.”

Several filmmakers have discussed using AI to de-age stars, typically for the flashback sequences of their films (whether this becomes a way for ageing stars—which Indian cinema has a lot of—to further prolong their careers as de-ageing tech becomes more sophisticated is worth keeping an eye on). Lokesh Kanagaraj, director of Tamil blockbusters Leo and Vikram, admitted to using AI to create a young Rajinikanth’s voice for Coolie’s flashbacks, the 2025 film’s most thrilling sequences. Director Guhan Senniappan similarly discussed using AI to de-age Sathyaraj for their 2024 Tamil film Weapon. As did Rekhachithram, one of 2025’s most acclaimed Malayalam films, which featured a cameo from an AI-enabled, digitally de-aged Mammootty. Its director, Jofin T. Chacko, said in an interview last year to online news portal Onmanorama that he hadn’t planned on using AI when work started on the film five years earlier, but the technology had since leaped ahead. “Recreating a specific look of Mammootty from a particular movie was challenging,” Andrew Jacob D’Cruz of Mindstein Studios, which led the de-ageing process, said in an interview to The Cue, a digital news platform. “It took us four-five months to recreate Mammootty’s face properly. Since AI-generated results vary with each attempt, we had to create four different versions of his face for every shot.”

Several filmmakers and producers say that post-production is where AI is currently having the most impact within mainstream moviemaking. One producer said that the often-costly process of patchwork, reshoots and voiceovers is now far less expensive and time-consuming thanks to AI. The producer added that top streaming platforms have thus far had a firm “no AI” policy for live-action filmmaking, but are now revisiting these policies as long as no AI is used for “human interaction” scenes between characters as opposed to more generic elements like establishing shots using AI-generating video tools like Sora and Adobe Premier Pro.

Colourist and film producer Sidharth Meer (Jubilee, Sabar Bonda) says that within the world of colour correction, the previously time-consuming task of face-tracking across thousands of shots is now much less demanding as a result of AI tools. “What would have taken a month or so to do can now be tackled in three days,” he says. AI-assisted dubbing and “localisation” is another increasingly common practice, where films can now be dubbed in other languages with the actors’ mouth movements matching the various languages with AI software. War 2 (2025) reportedly used the technology for the film’s Telugu dub. A person who worked on the film confirmed that AI was even used to “completely reconstruct” the voice of a young actor in his debut film—a love story for a major production house last year—“as a result of the actor’s speech impediment”.

AI is coming for trailers, too. Sahil Kajale, creative director at visual promotion agency Warriors Touch, has worked on several trailers, including Trapped , Omerta, Sacred Games and Raman Raghav 2.0. Kajale says that while AI hasn’t quite hit the stage of generating believable video footage, it has come for music, which is an essential part of deciding the fate of a trailer, and now risks becoming “supremely generic”. “Producers are trying to save every penny that they can at every stage, and AI-generated tracks don’t take money because they are not licence-driven. It’s not composed by a human being, so it doesn’t involve having to pay somebody.”

THE ETHICAL MINEFIELD

There is a key distinction between using AI to help traditional live-action filmmaking and the new “genre” of entirely AI-generated films. Vijay Subramaniam, CEO of Collective Artists Network, the creative architects behind the Historyverse, says the company is on their way to becoming the world’s “first and biggest AI studio”. When the company first announced the Chiranjeevi Hanuman film on social media, it was met with online backlash from filmmakers like Anurag Kashyap and Vikramaditya Motwane. “Here is the man heading the @lifeatcollectiveartistsnetwork that represents artists, writers, directors, now producing a film made by AI,” Kashyap wrote. “So much for looking after and representing the interests of creators.” To the outcry from the filmmaking community, Subramaniam offers: “This whole ethical conversation is becoming a little over the top. I genuinely don’t understand the hue and cry about it. …AI is not here to be fought. AI is here to be made friends with. I turn a blind eye and a deaf ear towards people who are anti-growth.”

He also insists that the production process for AI films isn’t “10 dudes furiously typing prompts into computers”, but actually involves a full team of artists. “The process is absolutely no different than what it would look like for an animation film. You have a director, you have an editor, you have somebody who’s designing costumes, you have a production designer, you have a dialogue writer, you have a scriptwriter, you have an assistant direction team, you have graphics engineers, you have 3D modellers. It’s no different. It’s just that you’re taking away the element of physically having to shoot. Mahabharat: Ek Dharmayudh has a crew of about 110 people behind it…If you think people sitting behind a computer and typing a prompt saying ‘now Hanuman jumps’, that’s just not the way it’s done.”

As for the shaky visual output of the JioHotstar show, he says the focus on the ugly AI aesthetic is overblown because it’s only the trained eye that can spot it. “I think the AI-ness of it is what is causing strong reactions. But there are lots of slightly older audiences that are not even able to tell it’s AI,” Subramaniam says. What excites him about the technology is the ability to bring our sweeping epics to life in an affordable way. He says AI-generated films are “50-80%” cheaper than attempting these projects through live-action.

Vivek Anchalia, director of the 2025 AI film Naisha, suggests the cost saving might be even greater. “The crew size of a (live-action) small feature film might be closer to 100 people,” he says. “But on AI, your crew size could be 15-20 people.” He isn’t as concerned about the ethics of AI in fiction filmmaking as much as he is about its implications elsewhere, especially as the technology gets better and more believable. “The younger generation is seeing so much AI content. At some point, if you don’t know if what you’re seeing is real or not, what do you believe? What happens if it’s a non-fiction film? What happens if someone creates a clip of a riot happening and it isn’t real?” With so many big Indian films presenting distorted and provocative versions of history as fact, this could become a very real problem.

Editor Nitin Baid (Homebound, Gully Boy) feels we’re panicking too soon. “Right now, it’s the initial phase. We still haven’t been able to fully decipher the potential or even say that it’s really bad or really good.” He adds that even though AI is a stage where it can generate shots, it’s still not able to achieve visual consistency. “As soon as you want like five shots which are exactly like in the same quality… we still need humans, because (AI is) not giving a perfect result.”

Baid says he has no issue with AI-generated storytelling if it can produce something that’s worthwhile and emotionally affecting. “General audiences aren’t even thinking about all this. They’re going to buy a ticket if they feel it’s something they want to watch and if it’s a story well told, as long as it feels real and heartfelt.”

There is good reason to be wary of the rapid spread and adoption of AI within filmmaking. It’s often said that a film is born three times: in the writing, in the shooting, and finally in the editing. We may well be stepping into a world of a fourth stage, which allows a film to be reshaped, rewritten, and reborn using AI. Protection against deepfakes and the ethical ramifications of AI likenesses are also worryingly open issues at present. Its increasing use will reduce the reliance on imagination and creativity, which could lead to a devaluing of human talent. There’s also a danger of cinema dividing along the lines of big-budget AI-assisted “spectacle” films and smaller non-AI releases. Still, on The Streaming Show podcast last year, director and producer Karan Johar offered a hopeful perspective. “What is natural and pure and human is always going to exist. And we’re going to be able to spot the difference. AI can never make Saiyaara.”

Suchin Mehrotra is a film critic and journalist.