The Fading Allure of Sora 2

The initial rush was intoxicating, a digital dopamine hit that few new technologies manage to deliver.

Barely a month ago, the arrival of Sora 2, OpenAI’s latest foray into AI-generated video, felt like a genuine revolution.

For a brief, glorious period, it was the only app that mattered, turning my phone into a hotbed of comedic creation and battery depletion.

I, like many early adopters, was utterly hooked, burning through my workday — much to my editor’s likely chagrin — crafting absurd, hyper-realistic video snippets of friends caught in the most improbable situations.

Peter Kafka, a colleague, found himself immortalized in a red jumpsuit, channeling Michael Jackson, while another friend was digitally subjected to the indignity of a ska band performance.

The hilarity was immediate, visceral, and deeply personal.

But the intoxicating fizz of novelty, as it often does in the relentless churn of the tech world, has evaporated.

Six weeks on, Sora 2 sits largely untouched on my home screen, a digital relic of a fleeting obsession.

My friends, too, have gone silent, their timelines no longer peppered with the uncanny valley antics of their AI-generated doppelgängers.

The question isn’t whether Sora is a technological marvel – it undeniably is – but rather, why did its grip on our collective attention prove so fragile?

The answer, at least for me, lay squarely in what I termed the “friend graph.”

The true genius, and indeed the primary source of amusement, was the ability to create highly specific, often embarrassing, in-jokes featuring people I actually knew.

There’s a distinct comedic difference between a generic video of a famous historical figure on a skateboard ramp and a bespoke clip of your friend Matt experiencing an unfortunate gastrointestinal incident at a rodeo.

The latter, fabricated by Sora, felt uniquely mine, a shared laugh born from real-world familiarity translated into digital absurdity.

This was where Sora truly outshone the ambient, “cool-looking” AI videos that platforms like Meta’s Vibes offered; it tapped into the deeply human desire for shared, personal humor.

However, this crucial “friend graph” proved to be a closed loop, never expanding beyond a handful of initial tech-savvy enthusiasts.

My friends outside the media and tech bubble, the very people whose reactions would have amplified the social fun, remained stubbornly on the sidelines.

The idea of granting an AI the ability to manipulate their likeness, to put their face into any conceivable scenario, was met not with laughter, but with genuine horror.

And who can blame them?

The underlying anxieties about digital identity, consent, and privacy are not easily assuaged by the promise of a silly video. For more on these topics, see IBM on privacy issues and AI in digital identity security.

Indeed, the privacy concerns extend beyond mere discomfort.

A darker, more unsettling dimension of Sora’s open-ended creative potential quickly surfaced.

I, and other women on the platform, began encountering what appeared to be fetish content, a stark reminder of the internet’s capacity to weaponize any tool for niche, often disturbing, purposes.

Witnessing an account dedicated solely to creating videos of women in tuxedos, only for someone to ritually rip off their boutonniere, was a stark, if bizarre, lesson in the internet’s boundless capacity for unexpected fetishes.

It’s a sobering counterpoint to the innocent fun, highlighting the ethical tightrope walk inherent in generative AI. See more on generative AI ethics here.

OpenAI, perhaps sensing a waning engagement, recently dropped the invite code requirement, opening Sora to the public in several key markets.

Yet, the news was met with a resounding shrug from my circle.

The initial buzz has dissipated, and the casual user base seems uninterested in a belated invitation to a party that, for many, has already peaked.

This isn’t to say Sora is a flop.

Far from it.

Its technical prowess is undeniable, and there’s a vast global audience eager for a high-quality, easy-to-use, and crucially, free AI video generation tool.

These users are likely finding myriad applications beyond “sending jokes to friends” – perhaps for content creation, marketing, or other yet-to-be-discovered professional uses. For insights into content building, check this analysis.

The irony is that while this broader adoption might validate Sora’s utility, it simultaneously presents a formidable challenge for OpenAI.

Running such a computationally intensive service for free comes at an astronomical cost.

Forbes’ back-of-the-napkin estimates suggest a staggering $15 million per day to keep Sora humming, a figure that casts a long shadow over its long-term viability as a free public utility.

The future of Sora remains a fascinating, if uncertain, question mark.

Will it evolve into a sophisticated, paid professional tool, shedding its ephemeral social novelty for serious utility?

Or will it find a way to reignite the spark of communal creativity, perhaps by addressing the privacy concerns that stifled its social expansion?

For now, the app represents a poignant case study in the lifecycle of digital novelty: a meteoric rise, a thrilling but brief reign, and then, for many, a quiet fade into the background.

As for me, the brief, intense affair with AI-generated shenanigans has concluded, and I’m thankfully back to wasting my time on my phone in more traditional, and perhaps less existentially complex, ways.