Google unveils Veo 2, rivals OpenAI in AI video generation

Google has unveiled Veo 2, its latest AI video generation model, just days after OpenAI released its own model, Sora, to the public.

The competition between Google and OpenAI continues to intensify. Following OpenAI’s launch of ChatGPT Search as a challenge to Google’s core search business, Google has responded with a significant move in the battle for dominance in AI-driven text and video generation.

According to a blog post by Google, Veo 2 produces highly detailed, realistic videos across diverse styles and subjects. The model excels in capturing the intricacies of real-world physics, human movement and expressions, enhancing overall realism. It also understands cinematographic language, allowing users to specify genres, lenses and cinematic effects to generate videos up to 4K resolution, lasting several minutes.


As per Google, unlike many video models that often “hallucinate” details, such as extra fingers or unexpected objects, Veo 2 significantly reduces these errors, delivering more accurate outputs.

Keeping safety in mind Veo 2 outputs include an invisible SynthID watermark that helps identify them as AI-generated, helping reduce the chances of misinformation and misattribution.

Access to Veo 2 is currently restricted to Google Labs' VideoFX. New users can sign up for the waitlist on Google Labs for access. Google also plans to integrate Veo 2 with YouTube Shorts and other products in 2025, signalling its commitment to AI-driven content creation in the future.