Artificially generated videos are on the cusp of becoming far more convincing, a development that carries both exciting creative potential and troubling societal implications. On Tuesday, OpenAI officially revealed the release of *Sora 2*, the newest and most advanced version of its flagship system dedicated to the generation of both audio and video content. In conjunction with this high-profile product launch, the company also introduced a complementary social application specifically designed to showcase these machine-created clips. This new app presents users with a continuous, scrollable feed of surreal, algorithmically crafted videos — a stream of content that, while engaging and impressive, also raises concerns about the degree to which it may saturate and disorient human perception.
To emphasize the capabilities of the new model, OpenAI published an announcement video which — according to the company — was produced entirely by Sora 2 itself. In this promotional material, a digitally fabricated version of OpenAI’s CEO, Sam Altman, dramatically described the software as “the most powerful imagination engine ever built.” The principal advancement emphasized in this latest iteration revolves around what OpenAI calls *world simulation*. This ambition entails enabling the model to reproduce the physical properties and constraints of the natural world, including the adherence to gravity, inertia, and other rules of motion. For example, the company presented demonstrations in which human figures within generated scenes exhibited much more lifelike and physically plausible movements than with previous AI video tools. Although the developers concede that imperfections and glitches remain, they confidently maintain that Sora 2 represents a significant step forward, achieving a much more reliable compliance with the principles of physics. Beyond improved realism, the model now also demonstrates sophisticated comprehension of elaborate, detailed prompts and is capable of generating multiple distinct shots from a single textual input, thereby broadening its narrative versatility.
Parallel to the launch of the upgraded model, OpenAI unveiled the *Sora app*, a social platform built to function as a dedicated showcase for AI-generated video content. As was speculated prior to launch and reported by outlets such as *Wired*, the app is designed in the style of modern social media feeds, delivering an endless vertical scroll of tailored videos curated through a recommendation algorithm. In an effort to make the experience more personal and potentially more uncanny, OpenAI created a feature called “cameo.” Through this option, users can insert themselves into generated videos, but only after filming a short verification clip to authenticate their identity and capture their facial likeness. Once this verification is complete, other participants in the community can incorporate that likeness into their own creations — subject, of course, to permissions granted by the original user. The effect is both intriguing and somewhat unsettling, evoking questions about the boundaries of digital identity.
In its public communications, OpenAI has taken pains to emphasize that users retain ultimate control over their likeness. The company affirms that cameo participation is opt-in and adjustable; individuals decide explicitly who can access their digital self, and they maintain the ability to revoke or remove any cameo-containing videos, even drafts created by different users. Moreover, participants with cameos embedded in other people’s content will have visibility into those creations at all times, helping them monitor and manage how their likeness is used.
The firm insists that both Sora 2 and its accompanying social application are being released with a framework of responsibility in mind. For example, OpenAI asserts that the design of the app prioritizes creative activity rather than passive consumption. The system is not optimized to maximize time spent scrolling the feed, a common criticism leveled at mainstream social media, but instead to encourage artistic exploration and the generation of new works. Still, one could argue that even creating within the app contributes significantly to time spent inside its ecosystem, meaning the distinction may be subtler than advertised. Importantly, the company pledges stricter safeguards for younger audiences: teenage users will encounter daily limits on the number of videos they can view, along with restrictions on how their verified likeness may be incorporated, in an effort to reduce potential harms and exploitation.
With respect to business strategy, OpenAI currently insists that monetization is not the immediate priority. Nevertheless, the company has already outlined its preliminary revenue pathway. Specifically, it plans to allow users to pay a fee to generate additional videos during periods when demand exceeds computing availability. The language used suggests an intent to maintain transparency, promising that any future shifts in monetization strategy will be openly communicated. However, it is difficult to ignore the possibility of advertising or other commercial models eventually surfacing within the social feed, even if those options remain unattributed or ambiguously described at present.
For now, access to the Sora social app is limited: it is available only by invitation, restricted initially to iOS users residing in the United States and Canada, though OpenAI promises international availability will follow. At launch, the app is offered free of charge. Regarding the Sora 2 model itself, it is accessible within the premium $200 per month *ChatGPT Pro* tier, aimed at users willing to pay for early access to the cutting edge of generative video technology.
In sum, the unveiling of Sora 2 and its accompanying app marks not only a technical leap forward in the realism of AI-generated audiovisual material but also a social experiment in distributing such creations at scale. The result is a phenomenon that blurs the line between artistic tool and viral entertainment feed, heralding a new chapter in the story of artificial creativity.
Sourse: https://gizmodo.com/openai-officially-launches-video-generator-sora-2-now-with-social-feed-2000665823
