Is Sora the Beginning of the End for OpenAI? (calnewport.com)

OpenAI has released Sora 2, a text-to-video model that can generate surprisingly realistic—and often absurd or troubling—short clips from simple prompts, and simultaneously launched a TikTok-like iOS app, Sora, that lets users create and consume algorithmically curated AI videos. The app removes much of the last vestige of human effort from viral video creation, producing everything from silly stunts to ethically fraught deepfakes; OpenAI currently gates generation behind paid tiers (ChatGPT Plus: $20/month for up to 50 low-res videos; $200/month for higher-volume/higher-res), reflecting the high backend cost of video synthesis. Technically this signals a major step in multimodal generative models becoming consumer-facing and real-time, but the release is significant for other reasons: it highlights content moderation and safety challenges, the economics of serving compute-heavy media (competing poorly on cost with ad-driven platforms like TikTok), and a strategic pivot by OpenAI from grand visions of imminent economic transformation toward monetizing attention and novelty. The move raises questions about whether current AI capabilities are being scaled into meaningful societal change or simply repackaged as entertainment—and whether a company that once warned about existential shifts is now chasing short-form engagement and paid perks.