OpenAI’s Sora 2 lets users insert themselves into AI videos with sound (arstechnica.com)

OpenAI unveiled Sora 2, the company’s second‑generation video-synthesis model that for the first time produces synchronized dialogue, sound effects and immersive background soundscapes alongside generated video. OpenAI also launched an iOS app that lets users insert themselves into AI‑generated clips via “cameos.” The company showcased Sora 2 with a photorealistic (if slightly off‑pitch) Sam Altman speaking against fantastical backdrops — from a ride‑on duck race to a glowing mushroom garden — highlighting multi‑shot coherency and stylistic variety. Technically, Sora 2 improves visual consistency and physical accuracy versus earlier releases, claiming to simulate complex motions (Olympic gymnastics, triple Axels) and avoid previous failure modes like object teleportation — e.g., a missed basketball now rebounds realistically. The model can follow more complex, multi‑shot instructions and synthesize richer audio than Sora 1, joining peers such as Google’s Veo 3 and Alibaba’s Wan 2.5 that already generate synchronized audio. For the AI/ML community, Sora 2 signals a step toward production‑ready synthetic video with integrated sound, accelerating creative workflows but also amplifying deepfake and policy challenges around consent, detection and responsible deployment.