ADVERTISEMENT
When Netflix rolled out The Eternaut, an Argentine sci-fi series, in July 2025, viewers were caught in the spectacle of a collapsing Buenos Aires building. What they didn’t know at first glance was that the scene wasn’t crafted by traditional special effects- it was powered by generative AI.
Co-CEO Ted Sarandos revealed that the sequence, which would have been prohibitively expensive with conventional tools, was created 10 times faster and at a fraction of the cost using AI-generated visuals. “The creators were thrilled with the result,” he said. For Netflix, which reported a 16% revenue rise to $11 billion in the same quarter, the experiment was more than just a flashy tech showcase. It was a signal: AI isn’t coming to entertainment- it’s already here.
But the move also reignited anxieties in Hollywood, where AI was one of the flashpoints of the 2023 actors’ strike. From fears of stolen creative labor to threats of mass job displacement, the industry remains split between awe and alarm.
In March 2025, Amazon’s Prime Video announced AI-aided dubbing on select international titles, starting with Spanish-language animated films like El Cid: La Leyenda and Mi Mamá Lora.
“AI-aided dubbing is only available on titles that do not have dubbing support, and we are eager to explore a new way to make series and movies more accessible and enjoyable,” said Raf Soltanovich, VP of Technology at Prime Video and Amazon MGM Studios.
Prime Video’s AI experiments don’t stop at dubbing. The platform now uses large language models (LLMs) to simplify verbose synopses into crisp, engaging summaries, making content discovery faster for viewers. The company insists on a hybrid approach—pairing AI with human oversight to maintain quality.
Closer home, JioHotstar is deploying AI for customer-facing experiences. In August, Akash Ambani unveiled RIYA, a voice-enabled assistant that helps users discover shows with natural queries. Instead of scrolling endlessly, a viewer can simply ask, “Show me romantic comedies in Tamil,” and RIYA responds.
For India’s crowded OTT market, where hyper-personalization and regional content are key growth drivers, tools like RIYA mark a shift toward deeper audience engagement.
Storyboarding and Subtitles to the Regional Frontier
Aparna Ramachandran, Head of Digital Originals at Balaji Telefilms, describes AI as a double-edged sword: a speed enabler in production, but a potential creative trap.
“On the positive side, AI really helps speed things up. In production, it supports early steps like script ideas, storyboarding, or even rough visualizations. In post-production, it makes a big difference with subtitles, translations, dubbing, and editing,” she says.
Balaji is already using generative AI in pre-production for visualization, research, and VFX concept work, and is eyeing deeper use in post. “Workflows that were taking weeks before can now potentially be finished in hours,” Ramachandran explains. But she admits the technology is far from foolproof. “Human intervention is required because AI tends to hallucinate and give responses that might not be show-specific.”
She stresses that over-optimization risks dulling creativity: “If everything is optimized by algorithms, content could start to feel the same, with less room for bold, original ideas.”
Interestingly, Balaji has gone beyond pilots to invest in AI-driven show production systems that can generate full episodes and visual assets—turning them into original IP.
“This strengthens both cost efficiency and long-term creative capability,” she says. Yet, she is candid about economics: so far, AI tool subscriptions are adding to costs, not reducing them. “Shows made completely in AI may be cheap, but for high-quality productions, it’s too early to discuss cost savings.”
For Kaushik Das, founder & CEO of AAO NXT- Odisha’s first independent OTT platform- AI is less about cutting corners and more about breaking language barriers.
“AI is changing OTT across production, discovery and operations. Positively, it speeds up routine post-production tasks, helps with localization- transcripts, subtitles, dubbing- improves personalized recommendations and enables faster promo testing,” he says.
AAO NXT leans on AI for automated transcription, subtitle checks, audio clean-up, and first-pass post-production tools. The payoff: faster creative iterations and more bandwidth for human teams to focus on storytelling.
“AI is also helping us expand Odia content to wider audiences, by enabling subtitling and dubbing in multiple languages,” Das explains.
On the operations side, AI helps with piracy detection and content security, safeguarding creator rights.
But Das draws a red line at replacing human imagination: “We treat AI as a tool to improve efficiency and reach while keeping editorial control and creator consent central. Future investments will follow the same principle: practical, ethical and aligned with our mission to grow regional storytelling.”
The Bigger Picture: Speed vs Soul
As OTT giants global and local rush to deploy AI, the benefits are clear—cost savings, faster turnaround, and wider accessibility. Filmmaker Siddharth Roy Kapur earlier admitted to Storyboard18 that AI “sharpens human minds by making pre-production, VFX and character design more efficient.”
Yet the ethical flashpoints are multiplying. When Eros International re-released Raanjhanaa with an AI-altered ending where the hero lives, its director Aanand L Rai blasted the move as “deeply disrespectful” and a “reckless takeover that strips the work of its soul.”
For its part, in a recent blog post, Netflix acknowledged the sensitivities around the deployment of AI and the rapidly evolving legal environment, while emphasising that generative AI tools- capable of producing video, sound, text and images- can serve as “valuable creative aids” when applied transparently and responsibly.
The newly issued guidelines set out conditions for creators wishing to integrate AI into their work. Among the key stipulations are that AI outputs must not replicate or substantially recreate identifiable traits of copyrighted or unowned material, and that the tools used should not store, reuse or train on production inputs or outputs.