Let’s go back to February 2024.
At the time, text-to-image models dominated the AI conversation. Tools like Midjourney and DALL·E were flooding social media with surreal images created from a few lines of text. Video generation tools already existed. Platforms like Runway had begun experimenting with AI-generated clips. But when OpenAI introduced Sora on February 16, 2024, the scale and quality of its demonstration changed the conversation.
Sora can generate short, cinematic videos from simple prompts. Complex camera movements, realistic lighting, and layered scenes and characters can emerge from a few sentences typed into a prompt box. Within hours of the announcement, the internet began testing the limits of what this model could produce.
Users were encouraged to submit prompts and watch them turn into videos. The results ranged from the imaginative to the bizarre. A bicycle race across the ocean with animals as competitors. Two dogs hosting a podcast on top of a mountain. A grandmother turned social media influencer teaches viewers how to make homemade gnocchi. A woolly mammoth roams the snow-covered mountains.
https://t.co/qbj02M4ng8pic.twitter.com/EvngqF2ZIX
— Sam Altman (@sama) February 15, 2024
These clips were less about storytelling and more about demonstrating what the models could do. But they hinted at something even bigger. This suggests that video production, historically one of the advertising industry’s most resource-intensive formats, may soon become much more accessible to marketers and advertisers.
Brands immediately began experimenting with this technology. One notable example is Toys R Us, which used Sola to create short brand films that recreated the childhood imagination of founder Charles Lazarus. The video depicts a dream-like sequence in which young Lazarus imagines a world filled with toys before the brand is finally born.
At the time, many experts suggested that tools like Sora had the potential to democratize video production and reimagine advertising storytelling by reducing the cost and complexity of creating visual narratives.
Two years later, the story took a different turn.
From breakthrough to shutdown
In March 2026, OpenAI announced that it would be shutting down Sora, including a standalone app released in September 2025 that allows users to generate and share AI videos. The move also ended a proposed $1 billion partnership between OpenAI and Disney that would have allowed Sora to generate videos using characters from Disney’s portfolio.
The closure comes as OpenAI shifts its focus to enterprise tools, coding products and research related to artificial general intelligence, Reuters reported. The compute-intensive infrastructure required to run consumer video generation platforms is reportedly taxing internal resources.
For many observers, this decision did not necessarily mark the end of produced video, but rather the beginning of a more practical phase.
Rajni Daswani, chief growth officer at SoCheers, says the economics of tools like Sora remain difficult to justify at scale.
“The ROI on tools like Sora is notoriously skewed, making them more of a liability than an asset. The infrastructure is simply not ready to handle this billion-user platform,” she says. “For those of us in advertising, this is essentially a coming-of-age period, moving away from the era of experimental wizardry and into a phase where the industry is looking at ways to make these tools sustainable and safe for brands to use over time.”
Vishal Kumar, vice president of digital at Madison Media, believes the closure highlights the speed at which the AI video ecosystem is evolving.
“At the very least, Sora’s closure highlights how competitive and rapidly evolving this field is. New models are already far exceeding early benchmarks in quality, format, and speed, confirming that the trajectory of generative video is accelerating rather than stalling,” he says.
Jyoti Chugh Bhatia, group director at Gozoop Creative, sees this development as part of a broader reality check for the industry.
“Honestly, this feels less like a setback and more like a reality check for where generative video actually stands today. The promise of faster production, lower costs, and more room for experimentation is still exciting, but we’re not yet at a point where core storytelling can take over,” she says.
Sora’s own growth reflected both the excitement and the limitations of space. The standalone app saw strong early adoption, reaching over 1 million downloads within days of launch and peaking at over 3 million downloads worldwide. But maintaining engagement has proven more difficult, with the platform facing increased scrutiny over copyright, deepfakes, and celebrity abuse.
AI Video in Marketing: Combining Experimentation and Efficiency
Despite Sora’s closure, generative video tools are already starting to impact how brands approach their production workflows.
Rather than completely replacing traditional production, agencies are increasingly using AI video to accelerate parts of the creative process.
“Today, brands are already meaningfully using AI video within their production workflows, not to replace big creative ideas, but to unlock scale through faster versioning, resizing, and platform-specific adaptation,” Kumar says.
He added that as audience targeting becomes more restricted across digital platforms, the ability to create multiple creative variations quickly is becoming a performance advantage for marketers.
Daswani points out that agencies are already seeing measurable time savings in early creative work.
“At the agency level, we are already seeing AI reduce storyboarding and previs time by 40-50%,” she says. “But the cost has shifted from production staff to rapid engineering and extensive post-production tweaks. And you still need a human eye to make sure the physics aren’t messed up.”
She adds that real efficiency gains are likely to emerge in the intermediate stages of a marketing campaign, especially when expanding content variation across different formats and platforms.
However, the increasing adoption of AI-generated video has also raised new concerns. The same technology that allows marketers to quickly create content can also enable deepfakes, misinformation, and intellectual property misuse.
Ironically, the day before announcing the shutdown, OpenAI published a blog outlining the safety mechanisms embedded in Sora, including watermarking, a content management system, and guardrails to prevent the generation of harmful or misleading content.
Even with these protections, concerns about authenticity continue to shape how marketers approach AI-driven content.
Trust, creators, and the limits of automation
As AI-generated media becomes more sophisticated, trust has emerged as one of the biggest challenges.
Research shows that viewers still have greater trust in human-driven content. According to Rukam Capital’s report, Beyond Metros: The Real Story of Bharat’s Next 500 Million, digital discovery in India’s non-metros is primarily video-first and social-driven, with creator recommendations influencing 23% of consumers.
When AI is involved, the level of trust also changes. According to the Nielsen survey, while 87% of respondents expressed moderate to high trust in AI tools overall, awareness of the role of AI in content and advertising has declined to 69%. Discomfort increases when viewers realize that the content was generated by AI. About 55% say they are uncomfortable with websites that rely heavily on AI-generated articles, and 48% say they don’t trust brands that advertise on such platforms.
This tension between automation and reliability is shaping how brands deploy AI video tools today.
Bhatia said most marketers are currently using technology for small, experimental applications rather than flagship campaigns.
“Right now, most brands are using it for quick content, testing ideas and personalization rather than big, impactful movies,” she says. “There are also real concerns about deepfakes, copyright and misinformation, making marketers a little more cautious.”
Kumar points out that these concerns reinforce the value of creator-driven storytelling.
He said rising concerns about deepfakes and misinformation have strengthened the role of influencers and creators in helping their audiences recognize their authenticity.
“Looking to the future, AI-generated video will become an essential infrastructure for marketing, increasing creativity and efficiency, but authentic human insight and emotion will remain irreplaceable,” he says.
Licensing battles and a divided future
Given the intellectual property concerns in shaping the generated video debate, Disney was trying to address this issue by working with OpenAI, rather than against it. The proposed deal would have allowed Sora to generate content using Disney characters under managed licensing agreements.
However, the closure of Sora ended this partnership before it could materialize.
“The collapse of the Sora-Disney deal is a huge warning that licensing and enforcement are incredibly tricky,” Daswani said. “Brands are rightly afraid that their IP will suffer some sort of AI failure, or worse, a controversial deepfake illusion from which there is no going back.”
He added that the shutdown may force marketers to rethink how they approach AI tools.
Rather than relying on a single platform, the ecosystem can be fragmented into specialized tools designed for different types of production. For example, platforms like Runway are used for professional-level visual assets, while platforms like Pika are emerging for social-first content creation.
Daswani said this fragmentation reflects broader changes in the evolution of technology.
“The withdrawal of OpenAI forced us to stop waiting for a single tool and start looking for specialized niche ecosystems that actually deliver functionality,” she says. Sora’s rise and closure happened in less than two years. During this time, we demonstrated how AI can quickly reshape creative possibilities, while also revealing the challenges that come with rapid technological change.
For the marketing and advertising industry, the lesson may not be about the failure of one tool. Rather, it highlights the gap between technological capacity and the legal, economic and ethical framework needed to sustain it.
