What AI can now do·May 20, 2025·Media

Veo 3 generates video with synchronized dialogue; Sora 2 puts it in a free app

Google DeepMind; OpenAI

Veo 3 was Google's first video model to generate audio with the picture, including character dialogue, for US Ultra subscribers; OpenAI's Sora 2 (Sept 30, 2025) added synchronized sound and 'cameos' of real people in a free, invite-based social app.

How far it has travelled

From a demonstration to real life, in five steps. Empty steps are the point: a demonstration is not a deployment.

Demonstratedshown once, by its developer or a researcher
yes · May 20, 2025 · Veo 3 generated video with background sound and character dialogue 'for the first time' (Google's framing); outputs carry SynthID watermarks (10B+ items watermarked since 2023). source
Reproducedshown again by someone else
yes · Sep 30, 2025 · Sora 2: synchronized dialogue and sound effects, multi-shot consistency, and user 'characters' made from a short recorded clip; OpenAI said the model still makes mistakes. source
Accessibleavailable to the public or to customers
yes · Sep 30, 2025 · Veo 3 via the Gemini app and Flow for US Ultra subscribers and via Vertex AI; the Sora app was free at launch in the US and Canada (invite-based); the Sora product was discontinued on Apr 26, 2026. source
Reliableworks most of the time on a defined task
partial · Sep 30, 2025 · Physics and consistency improved but 'the model still makes mistakes'; watermark-based provenance depends on the generator cooperating and is absent on third-party tools. source
Real consequenceshas changed something in the world, for good or ill
yes · Oct 9, 2026 · Sora 2 produced deepfakes of deceased public figures (opt-outs added); by Oct 9, 2026 the Wesleyan Media Project had counted 164+ AI-made US midterm ads backed by $80M+, with 31 states requiring labels and no federal law; Texas saw a first criminal complaint under its election-deepfake law. source

What this could change for you

Realistic video of anyone saying anything can be made on a phone for free, so video now needs provenance before it can be believed.

Incidents it made possible

Sources

primary source, second source.

A milestone is a dated, sourced result that moved what AI can do; a stage is upgraded only when a source shows it (a reproduction, a public release, production evidence, a documented consequence). “Unknown” is the honest default.