OpenAI just gave up on its Sora AI video generator
openai sora
| Source: The Verge | Original article
OpenAI announced on Tuesday that it is pulling the plug on Sora, the text‑to‑video model it unveiled at the end of 2024. The brief statement, “We’re saying goodbye to Sora,” marks the end of a product that generated a wave of excitement for its ability to produce minute‑long, photorealistic clips from a single prompt, but also sparked controversy over its massive compute demands and the legal fallout from a collapsed partnership with Disney.
The shutdown follows a string of reports from earlier this week that OpenAI was already winding down Sora’s support and cancelling the multi‑million‑dollar deal with Disney that had promised exclusive content rights. As we reported on 25 March, the company had begun “killing” Sora and the Disney agreement fell apart amid concerns that the technology could blur the line between real and synthetic media. The decision now appears final, with the service being removed from the API dashboard and existing user credits slated for refund.
Why it matters is twofold. First, Sora’s demise underscores the practical limits of current AI video generation: rendering a single minute of high‑definition footage can consume more GPU power than many of OpenAI’s other flagship models, making it financially unsustainable at scale. Second, the episode highlights growing regulatory and reputational pressure on AI firms to curb tools that could be weaponised for deep‑fake propaganda or copyright infringement.
What to watch next is OpenAI’s strategic pivot. The company is likely to redirect the compute budget earmarked for Sora toward its next‑generation text‑and‑image models, while competitors such as Runway, Google DeepMind and Meta’s Make‑It‑Real may try to capture the vacated market segment. Observers will also be keen to see whether OpenAI offers a lighter‑weight video prototype in the future, and how regulators respond to the broader implications of AI‑generated media.
Sources
Back to AIPULSEN