Just this week, we, the two writers behind this piece, found ourselves captivated for hours, endlessly scrolling through a stream of captivating short videos. What made them so compelling? Each one featured us, the authors, in wildly imaginative and often absurd situations.
Imagine this: a nine-second, incredibly lifelike video of us fearlessly skydiving, huge grins plastered on our faces, with delicious pizzas serving as our parachutes.
{“visibleWhen”:[“before-play”]}
{“url”:”https://static01.nyt.com/images/2025/10/02/arts/OPENAI-SOCIAL-PARACHUTE_COVER/OPENAI-SOCIAL-PARACHUTE_COVER-verticalTwoByThree735.jpg”}
{“controls”:[“fullscreen”,”mute”,”share”,”scrubber”]}
Another clip showed Eli, our co-author, smashing a game-winning home run in a surreal baseball stadium, with robots cheering from the stands.
{“visibleWhen”:[“before-play”]}
{“url”:”https://static01.nyt.com/images/2025/10/02/arts/OPENAI-SOCIAL-BASEBALL-COVER/OPENAI-SOCIAL-BASEBALL-COVER-verticalTwoByThree735.jpg”}
{“controls”:[“fullscreen”,”mute”,”share”,”scrubber”]}
And in perhaps the most bizarre, Mike found himself locked in a "Matrix"-esque showdown against none other than Ronald McDonald, their weapons of choice: flying cheeseburgers.
{“visibleWhen”:[“before-play”]}
{“url”:”https://static01.nyt.com/images/2025/10/02/arts/OPENAI-SOCIAL-RONALD-McDONALD/OPENAI-SOCIAL-RONALD-McDONALD-verticalTwoByThree735.jpg”}
{“controls”:[“fullscreen”,”mute”,”share”,”scrubber”]}
Eli’s reaction to the cheeseburger duel? "I’m genuinely blown away," he messaged Mike, quickly giving it a like. Meanwhile, Mike, clearly enjoying himself, continued sharing videos – featuring him ballroom dancing with his dog and even perched on a throne of rats – with other New York Times colleagues (all of whom found the clips slightly disturbing).
The platform behind these bizarre, hyper-realistic clips wasn’t any of the usual suspects like TikTok, Instagram Reels, or YouTube Shorts. Instead, it was Sora, a groundbreaking smartphone application from OpenAI. This app allows users to generate impressive short videos entirely through artificial intelligence. While Sora’s core technology was revealed last year, its newest iteration — boasting enhanced speed, power, and the ability to integrate your actual likeness from uploaded photos — launched this week to an exclusive, invitation-only group.
Within mere hours of using it, we realized Sora was far more than just another AI video creator. It functions as a sophisticated social network, almost a mirror image of TikTok, complete with a familiar interface, algorithm-driven video recommendations, and features for following and engaging with friends. The advanced AI at its core makes video production incredibly easy, offering users an endless capacity to generate AI-powered content.
Yet, this immense creative freedom comes with an unsettling side.
Early adopters of Sora quickly began churning out videos featuring copyrighted characters from popular culture. We encountered a surplus of "Rick and Morty" and Pikachu clips, to say the least. When Mike shared one of his Sora creations on Instagram, several friends genuinely questioned if it was truly him in the video, highlighting a disturbing blur between reality and AI-generated content.
Even more concerning is Sora’s potential to supercharge disinformation. The ability to rapidly generate convincing video likenesses of individuals could lead to the creation of highly realistic fake events, potentially inciting real-world reactions. While other AI video tools already pose this risk, Sora’s ease and accessibility could drastically amplify the problem.
It’s still early, and Sora’s long-term success isn’t guaranteed. However, OpenAI seems to have achieved what tech giants like Meta and X have been striving for: a widespread AI product that fosters sharing, encourages user-generated content, and drives consistent engagement with their platforms.
The competition to develop similar applications is rapidly intensifying. Just last week, Meta unveiled "Vibes," a social media feed integrated into its dedicated AI app, leveraging an AI video generator from the Midjourney startup. Not to be outdone, Google offers its own comparable product, Veo.
As the social internet evolves from simple text messages to shared photos and now billions of hours of video, tech leaders are convinced that AI video tools will be crucial in shaping the future of social media.
Rohan Sahai, OpenAI’s product lead for Sora, explained their strategy: "We believed the most effective way to introduce this technology to a broad audience was through a somewhat social medium. With such a rapidly changing technological landscape, including new content formats, our company’s core philosophy is to get our innovations into the hands of users."
(It’s worth noting that The New York Times has previously taken legal action against OpenAI and Microsoft, alleging copyright infringement of its news content in relation to their AI systems. Both companies have publicly refuted these claims.)
Sora’s fresh design strikingly resembles TikTok, even adopting the familiar "For You" label for its social feed. Users have the option to scan their own faces to create personalized avatars for their videos, or they can even use images of public figures like OpenAI’s CEO, Sam Altman. OpenAI has dubbed this feature "Cameos," a nod to the well-known app that lets fans purchase personalized video messages from celebrities.
Safety experts have voiced concerns that Sora, particularly its Cameos feature, could open doors to novel forms of misinformation and online scams. A prime example is a viral Sora video depicting a fabricated version of Sam Altman appearing to steal a graphics processing unit from a department store, complete with the look of security camera footage.
Rachel Tobac, CEO of cybersecurity startup SocialProof Security, highlighted the alarming ease of creating convincing deepfakes with Sora, stating, "It makes it really easy to create a believable deepfake in a way that we haven’t quite seen yet."
While Sora does enforce limitations on certain explicit and copyrighted content—allowing, for example, videos featuring "South Park" characters but not Batman or Superman—the onus is largely on rights holders. They must proactively opt out of their work being used in Sora by submitting a copyright disputes form individually for each case. Conversely, public figures have the option to explicitly grant permission for their images to be utilized by Sora.
Varun Shetty, OpenAI’s head of media partnerships, confirmed their approach in a statement: "We’ll collaborate with rights holders to prevent specific characters from appearing in Sora videos upon their request, and we will address all takedown notices promptly."
This week, as Sora-generated videos started appearing across platforms like X and TikTok, public reactions were varied, ranging from astonishment and amusement to outright revulsion. A primary concern is that Sora will contribute significantly to the phenomenon known as "slop"—a derogatory term for the overwhelming influx of meaningless, AI-generated videos inundating social media feeds.
The rise of AI-generated "slop" is already evident. Last July, an AI clip depicting a baby mistakenly piloting a 747 became one of YouTube’s most-viewed videos. More recently, an AI-created video of an elderly woman shattering a glass bridge with a boulder went viral on Facebook and X, leading to countless imitations.
Mr. Sahai of OpenAI posits that, much like other social platforms democratized creative tools, Sora will lead to a vast spectrum of content quality. He believes that while much will be produced, the most exceptional work will naturally gain prominence. He also pointed out that what might appear as "slop" to a general audience could be perfectly relevant and amusing inside a close-knit group of friends.
“One man’s slop is another man’s gold,” Mr. Sahai wisely observed.
In Hollywood, the past 36 hours have been consumed with alarm over Sora’s potential to enable users to easily replicate celebrity likenesses without compensation. A day following the app’s launch, WME, a prominent talent agency, circulated an internal memo to its agents, affirming their commitment to fiercely protect their clients’ intellectual property, a document reviewed by The Times confirmed.
The WME memo emphasized the urgent need for robust protections for artists and creators against AI models that exploit their intellectual property, name, image, and likeness. WME also stated that they had informed OpenAI that all of their clients were explicitly opting out of any inclusion of their likeness or intellectual property in Sora’s generated videos.
Despite the concerns, Sora’s widespread appeal was undeniable from the start. Neither of us had any prior experience in video creation. Yet, with just a basic idea, a couple of minutes of processing, and significant computing power, we effortlessly generated a video of Mike and Eli arm-wrestling for the unofficial title of "best tech reporter." (Eli emerged victorious, by the way).
However, not everyone found the results charming. After Mike presented his partner with an unsettlingly realistic Sora video of himself portraying the psychopathic character Anton Chigurh from the 2007 film adaptation of "No Country for Old Men," her response was direct and clear.
“Please never, ever show me this kind of video again,” she pleaded.
Nicole Sperling contributed reporting from Los Angeles.